A Lonely Way to Die: An examination of deaths in police custody

A Lonely Way to Die: An examination of deaths in police custody

Dissel, A. & Ngubeni, K. (1999). A Lonely Way to Die: An examination of deaths in police custody. Research report written for the Centre for the Study of Violence and Reconciliation, July.


Amanda Dissel & Kindiza Ngubeni

Research report written for the Centre for the Study of Violence and Reconciliation, July 1999.

Amanda Dissel is Manager of the Criminal Justice Programme at the Centre for the Study of Violence and Reconciliation.

Kindiza Ngubeni is a Senior Community Facilitator in the Criminal Justice Programme at the Centre for the Study of Violence and Reconciliation.


This report is based on research conducted by Amanda Dissel and Kindiza Ngubeni. We would also like to express our appreciation of the support given by David Bruce, who co-ordinated the work on behalf of the ICD, and for other members of the team working on this project.

We would like to thank Commissioner Fivas for granting us permission to conduct this research, and to all the members of the South African Police Service who assisted with this research, and who spoke freely to us during our interviews with them. We also thank all the members of the Independent Complaints Directorate (ICD) who assisted us with statistics and information, as well as those whom we interviewed. None of the interviewees are acknowledged directly in the text.

Finally we would like to express our appreciation to the Royal Danish Embassy for their support in this project. The contents of this report do not necessarily reflect the views or opinions of the Royal Danish Embassy.

Abbreviations Used

CAT:United Nations Convention Against Torture and Other Cruel Inhuman and Degrading Treatment or Punishment
CPF:Community Police Forum
CSVR:Centre for the study of Violence and Reconciliation
ICD:Independent Complaints Directorate
SAPS:South African Police Service


During South Africa's apartheid past, the police became internationally notorious for the high number of deaths occurring in police custody. Most of the cases which received attention in the past related to the death of political activists, and it was seldom that news was made of an 'ordinary criminal' who had died in custody. As a result, there is little information on the actual number of people who died in custody. Since the establishment of the Independent Complaints Directorate, which has oversight over serious actions on the part of the police, this issue has been taken more seriously. Despite Constitutional principles protecting human rights, the number of deaths in custody still appear to be shockingly high indicating that there are serious shortcomings in the way that such people are handled while in custody.

The Centre for the Study of Violence and Reconciliation was commissioned by the Independent Complaints Directorate to conduct research into several areas within its statutory mandate. One of these areas was an examination of the problem of deaths occurring in police custody.

This report is divided into three sections. Section One examines the incidence of deaths in police custody in South Africa, and includes a detailed analysis of cases of deaths in police custody in Gauteng that were reported to the Independent Complaints Directorate (ICD). Section Two looks at some of the issues involving the management of people in police custody, specifically focusing on those held in police cells and the treatment they receive at the hands of the police. Section Three contains recommendations for the police on improving the management of such people, and for the ICD on conducting investigation into deaths in police custody.

Section One: The incidence of deaths in police custody


In terms of Section 53 (2)(b) of the South African Police Service Act, No. 68 of 1995, the Independent Complaints Directorate 'shall mero motu or upon receipt of a complaint, investigate any death in police custody or as a result of police action'. Subsection 53(8) states that the National or Provincial Commissioner shall notify the directorate of all cases of death in police custody or as a result of police action. The ICD may also receive independent reports from family members of the deceased individual, or through members of the public or government.

The ICD classifies complaints received into different categories, with Class I constituting complaints alleging 'the death of a person in police custody or as a result of police action'. The ICD further distinguishes between deaths in police custody, and deaths as a result of police action.

This report focuses on those deaths which occur in police custody rather than deaths which are the result of police action1 and which occur outside of police custody.

The ICD has no written definition of when a death should be classified as having occurred in police custody or not. For the purposes of this report we have defined police custody as any situation where a person is under the control of the police as a result of having been arrested, or is under the control of the police for any other reason, or is being held in police cells.2 This includes, but is not exclusive of any situation where a person has been arrested or detained by the police. This may include situations where a person:

  • is held in police custody at a police station, in a police vehicle, or in hospital (usually under police guard), or in temporary police accommodation,
  • is held in a police cell or under police guard but who does not fall into the category of people who have been arrested on the basis of an alleged criminal offence. This includes convicted prisoners who are held in police cells or under police guard, and mentally ill or impaired people awaiting transfer to a mental institution,
  • has been stopped and searched or questioned by the police,
  • is at the police station or in a police vehicle whilst not in police detention, such as a victim, complainant, or witness,
  • is attempting to escape from the police after having been arrested or detained,
  • who is injured by parties other than the police, but who is arrested or detained by the police, and subsequently dies whilst under the police's control.

For our purposes, a death in custody does not include instances where the person is:

  • injured or killed while attempting to resist police's attempt to arrest or detain him or her.
  • who is killed or receives injuries at the hands of the police prior to his/her arrest.

Since its inception the ICD has been collating statistics on deaths in police custody. Although the SAPS have been recording such statistics, these are not such a reliable source, as there has been no standardised system for categorizing deaths in police custody, or as a result of police action.3 In this report, the analysis of deaths in custody is therefore based on statistics provided by the ICD starting from April 1997.

It is not certain to what extent the statistics collected by the ICD on deaths occurring in custody are reflective of all of the deaths occurring in custody. According to the ICD Annual report for 1997/1998,4 it was noted that '(a)lthough in general the police complied with their statutory obligation of reporting instances of deaths in police custody or as a result of police action … the reports were quite often received late … and in some instances the reporting was not done at all'. The report goes on to say that after a workshop with the police, this situation improved somewhat.

Table 1: Deaths in police custody recorded by the ICD from April 1997 to March 1998
ProvinceApril 1997 – March 1998April 1998 – March 1999Total number of deaths over two year period% of deaths
Northern Province136%74%204.7%
North West105%158%258.37%
Free State84%105%184.3%
Eastern Cape2913%137%4210.04%
Western Cape2411%2915%5312.67%
Northern Cape52%70.95%122.87%

According to the figures outlines in Table 1 above, deaths occurring in police custody declined 14% in the second twelve-month period. This decline can be attributed to a decline in the number of deaths in four provinces (Gauteng, Northern Province, KwaZulu-Natal and Eastern Cape). In contrast the number of deaths increased in the five other provinces (North West, Mpumalanga, Free State, Western Cape and Northern Cape).

The apparently substantial overall decline in the total number of deaths in police custody over the two year period therefore reflects a slightly confusing picture, with an apparent decline of 55% in one province (Eastern Cape), paralleled by an increase of up to 71% in another (Mpumalanga).

The highest number of deaths in police custody for the two-year period is recorded in Gauteng (127), followed by KwaZulu-Natal at 92. While the least amount of deaths in police custody occur in Northern Cape, Free State, Mpumalanga and Northern Province.

Figure 1: Deaths in police custody – April 1997 to March 1998, and April 1998 to March 1999

The rate at which deaths occur in police custody is calculated in relation to the population figures, and is expressed per 100,000 of the population5 (X = No. of deaths x 100,000/ population of province). These statistics are reflected in Figure 1 above. Gauteng recorded the highest rate of deaths in police custody per population in 97/98, but was overtaken by the Northen Cape in 98/99. In these two provinces, there was almost 1 death in police custody per 100,000 of the population. Despite having the lowest number of deaths over the two-year period (12), the Northern Cape recorded some of the highest rates when compared with its population size. The rate at which these deaths occurred was also high in the Western Cape and KwaZulu-Natal. In North West, Mpumalanga, and KwaZulu-Natal, the deaths occurred at approximately 1 death for every 200,000 of the population for the 1998/99-year. In respect of the same year, the deaths occurred at reduced frequency of 1 death for every 400,000 in Northern Province and Eastern Cape.

This indicates the treatment of people in custody is a greater problem in Gauteng and Northern Cape. While these figures provide some basis for comparison between provinces it would be more illuminating if they were compared to the number of people held in custody in each province during the 24-month period.6

Figure 2: Deaths as a result of police action, and deaths in custody during 1998

Figure 2 above illustrates the rate of deaths in custody and deaths as a result of police action per 100,000 of the population. This graph is calculated according to the figures of deaths in custody for the calendar year of 1998.

This graph illustrates that where there is a high rate of deaths that occurred in police custody, there tended also to be a high rate of deaths as a result of police action. This is particularly high in Gauteng, which reflected the highest rate of deaths in custody, and as a result of police action. The rate at which deaths occurred as a result of police action is two and a half times the frequency of deaths in custody. Deaths resulting from police action occurred at significantly higher rates (at three to four times the frequency) in Mpumalanga, KwaZulu-Natal and the Free State, and at almost six times the frequency of deaths in custody in the Eastern Cape. In Western Cape, which had relatively high rates of deaths in both categories, the deaths as a result of police action occurred at only 1.6 times the frequency as the deaths which occurred in police custody.

The Independent Complaints Directorate breaks down its figures on a provincial basis. The ICD statistics are also further broken down into 5 categories: deaths as a result of natural causes; suicide; deaths as a result of injuries occurring in custody; deaths as a result of injuries prior to custody; and deaths as a result of possible negligence. This study of deaths in custody looks at deaths that occurred from January to December 1998.

Table 2: Deaths in police custody from January to December 1998 as recorded by the ICD
ProvinceNatural CausesSuicideInjuries in CustodyInjuries Prior to CustodyPossible NegligenceTotal
Northern Province3111612
North West5501415
Free State6114012
Eastern Cape2502211
Western Cape31034626
Northern Cape210014
Total number of deaths5259254830204

Again over this period, the greatest number of deaths (60) are reflected as occurring in Gauteng,7 followed by KwaZulu-Natal at 53. These statistics are taken from overall provincial figures provided by the ICD. They do not provide a detailed analysis of the cause of death, and thus allow a limited ability to analyse causes of death in police custody.

During this period, the greatest number of deaths (29% of total number of deaths) occurred as a result of suicide. In Eastern Cape deaths resulting from suicide constitute 45% of deaths in custody, 40% in KwaZulu-Natal, and 38% in the Western Cape.

A high number of deaths (25%) are attributed to 'natural causes'. Whilst the highest number of these deaths was recorded in Gauteng, natural deaths constituted 56% of total deaths in police custody in Mpumalanga, and 50% in Free State and Northern Cape.

'Injuries which occur in custody' could be as a result of injuries sustained at the hands of the police, accidental injuries, or injuries which are self-inflicted, or inflicted by a third party. A similar deduction can be made in relation to 'injuries which occur prior to custody' – the agent at whose hands the injuries occur is unknown from these statistics. 'Injuries in custody' account for 12% of deaths, while 24% of deaths are attributed to 'injuries prior to custody'. Injuries that occur prior to custody account for 34% of the deaths in KwaZulu-Natal and 30% of the deaths in Gauteng.

Another issue in relation to these statistics is that the deaths are categorized shortly after the ICD has been notified of the death, and this category is not necessarily altered should some other cause of death become apparent during the course of the investigation.8 In addition, the investigation for many of the cases reflected in this table have not yet been completed, so the actual cause of death is not necessarily known.

A further issue is that the ICD records all deaths that are reported to it, whether or not they fall within the jurisdiction of the ICD. For example, if a death occurs in a prison, as opposed to a police cell, this is not a death that occurs in police custody at all, yet it would be reflected in the statistics. It could therefore be stated that these statistics are not a very accurate representation of deaths that occur in police custody.

In order to provide a more detailed analysis of deaths occurring in custody, the ICD needs to be able to obtain a greater understanding of the causes of deaths. The researchers have sought to do this through a detailed docket analysis in Gauteng.

Methodology for Gauteng Study

The CSVR considered it necessary to obtain a more detailed understanding of the causes and the nature of deaths that occur in police custody in order to make policy recommendations to the ICD. A detailed analysis of the ICD dockets concerning deaths in police custody in Gauteng during 1998 was undertaken.

It was felt that the deaths in police custody are only one aspect relating to the management of people in police custody, and that it is necessary to examine police practice, policy and attitudes towards the management of such people so that any systemic problems can be ascertained. Police cells were therefore visited at eleven police stations in Gauteng. Police officials in charge of detainees were interviewed, and in some cases, the Station Commissioner was interviewed. Four ICD investigators and those supervising investigators were also interviewed.

Gauteng Docket Analysis

CSVR requested the ICD at the Johannesburg and Pretoria offices to supply its researchers with the dockets relating to deaths that occurred in police custody in Gauteng police stations during the year 1998. As a result, the researchers were supplied with more than 60 dockets. Many of these were rejected by the researchers as they clearly documented cases where deaths had occurred during police action, or the deaths had occurred the previous year, or they had occurred in a province other than Gauteng. In all, forty-seven dockets which related to deaths in custody were examined.9 Each docket was scrutinized to ascertain various details relating to the deceased's arrest, date of detention, date and place of death, and information relating to the post-mortem report, if any.

The cases selected by the researchers were those that were classified by the ICD as deaths in custody. However, there were two cases included in the study which appeared to be cases of deaths in custody, but had been classified by the ICD as 'neglect of police officer's duty', and 'death as a result of police action' respectively. In the latter case, it was apparent that the person had been shot after arrest by the police, and whilst under the control of members of the SAPS. A list of all cases handled by the ICD during 1998 was supplied to the researcher by the Johannesburg office of the ICD. There were 12 cases classified as 'deaths in police custody' by the ICD, but which were not seen by the researchers. There was also one case (Josiah Rabotapi) which was registered by the ICD as a case of death as a result of police action, although other reports indicate that the deceased was in the custody of the police at the time of his death, and which the researchers did not have sight of.10 This indicates that there may have been other cases which were classified by the ICD as deaths other than in police custody, but which may have in fact been deaths that occurred in police custody. Furthermore this research does not represent all of the cases of deaths occurring in police custody for the relevant period, even in terms of the ICD records of such cases.

Analysis of the dockets proved to be difficult largely because investigations were not completed in 44% of the dockets. The circumstances surrounding the death, and the physical cause of death had not yet been ascertained. In the cases where matters had been referred to the Attorney-General, the outcome of his recommendation was not always available. A similar situation applied in respect of the inquests.

Table 3: Progress of cases being investigated by the Independent Complaints Directorate
State of ICD investigation /monitoring
Case sent to Attorney General111
Attorney General requested inquest2
Unsubstantiated: not involving police action18
ICD investigation not complete15
ICD recommended disciplinary action2
Closed: action justified1

Secondly, the amount of information contained within the dockets varied considerably. In the most extreme cases, the docket contained only the ICD form that the police complete in order to notify the ICD of the death. In other cases, there was no post mortem report, or contained only notes made by the ICD official who had attended the post mortem. In the majority of cases, no witness statements were contained in the file. These conditions also applied to many of the cases that had been closed, particularly those where the ICD had found no action on the part of the police. On the other hand, there were many files, which due to the extensive nature of the documents, indicated that a thorough examination by the ICD or the police had taken place.

Figure 3: Deaths in police custody according to category

Analysis of Deaths According to Category

Notwithstanding the difficulties, where the investigation is still pending this we have categorised the deaths examined according to the prima facie evidence available, or on the basis of the version supplied by the majority of the witnesses.

This study considered it more useful to allocate the deaths to different categories which provide greater details of the circumstances, if not the causes, of death. These categories have been derived from those used in the analysis of deaths in custody in a study conducted in the United Kingdom by Leigh et al,12 as this provides a more detailed indication of the circumstances surrounding the cause of death. Three broad categories were identified: deaths as a result of the deceased's own action (15 deaths recorded); deaths as a result of the deceased's medical condition (11 deaths recorded); and deaths where another person's actions may have been associated with the death (21 deaths recorded) of a total of 47 deaths. These broad categories have been further subdivided and analysed.

Figure 3 above indicates the spread of deaths in Gauteng according to each category.

The highest number of deaths occurred in the category of injuries inflicted by another person pre-custody. These accounted for 34% of the total number of deaths in the dockets studied. The category of 'deliberate self harm in custody' is the second largest category, at 25,5% of the total number of deaths. This reflects the high number of suicides occurring in police custody. Deaths resulting from the medical conditions of prisoners constitutes 21% of the total. As was noted in Table 2 above, deaths due to natural causes were disproportionately high in Gauteng compared with other provinces.

The CSVR undertook to respect the confidentiality of these cases. Therefore the names of the deceased, police stations, police officers or complainants have not been used in this study. Instead, the cases have been assigned alphabetical letters.

Deceased's Own Action

Table 4: Reflects deaths in police custody which were the result of the deceased's own action – 1998
Deceased's Own Action:
No other person's actions were associated with the death
Number of Deaths
Deliberate self harm (in custody)12
Deliberate self harm (pre-custody)1
Substance abuse (alcohol)1
Substance abuse (drugs)1
Deliberate Self Harm (in Custody)

Eleven of the twelve cases of 'deliberate self harm in custody' were instances of apparent suicide. Ten of these cases were the result of hanging. All except two of the deceased hanged themselves from the bars of the cell doors or windows of their police cells.

Two people hanged themselves in places other than police cells. One deceased hanged himself in the corridor between the holding cell of the court and the holding cells of the magistrate's courts cells. The other deceased apparently hanged himself from a 'monkey chain'13 in his hospital bed where he was receiving in-patient treatment.

The people who hanged themselves had used various instruments as ligatures, such as a monkey chain (1), shoelaces (1), leather belt (2), torn blanket (1), trousers, and the elastic of track suit pants (2), bandages covering the legs (1), and mutton cloth (1). The instrument used was not recorded in one case.

Indicators of Risk

The files were perused with a view to ascertaining whether there had been any warning indications that the deceased was likely to attempt to take his life. There was no direct warning indicated in any of the cases, and it did not appear that the ICD investigator or the police had asked this question. However, the circumstances of four of the cases could have alerted the authorities to potential problems. In two of these cases ('J' and 'L'), the deceased's had been 'acting strangely' when they had been taken into custody by the police. In J's case, the deceased was brought to the police because some people thought he was acting strangely and some people thought he was lost. Although he was brought to the police station earlier in the day, he was only put into a holding cell after members of the SAPS Trauma Centre had interviewed him. He was put into a holding cell at 15h20 and was found hanging at 15h57. He was noted to have a 'mentally unstable condition', and that his condition became worse as the day progressed. At one stage he tried to jump out of the window and reported that he was hearing noises in his head. The Police wanted to contact members of his family, but the deceased told them not to. He had been put into a holding cell 'for his own protection'. In L's case, the deceased was brought into the police station by his work colleagues who wanted to have him certified as a mental patient. The police placed him in a cell at 18h00, and when they returned to collect him at 06h30 the following day, they found that he had hanged himself.

In both of these cases, the deceased had not been arrested for any crime. In the former case, the police had not removed his personal effects because J had not been arrested. He hanged himself with his belt, whereas in L's case, the police had not even recorded his details in the cell register. In both cases, it seems that not even simple preventative measures had been taken by the police, such as removing articles of clothing which the suspects could use to harm themselves.

In the third case ('K'), the deceased was arrested for drunkenness after assaulting his wife. He had earlier been brought to the police station and held in the charge office to 'sober up'. He had been released and had returned some hours later. He began to 'make a fuss' and he was arrested for drunkenness. He was placed in the 'hard labour's cell'14 where he was to await transfer to another police station. When the police went to collect him, he was found hanging.

In the last case, that of 'H', the deceased had been arrested, and had attempted to escape while being booked into the cells. He had been apprehended and been booked in. He hanged himself with the elastic of tracksuit pants which were in the cell, apparently not his own.

Visits to Police Cells

The researchers recorded the time indicated between the last visit to the deceased and the time that he was found hanging. In two cases, these details had not been recorded. In three cases, the deceased had been visited at least half an hour before he was found hanging (in two of these cases the deceased had been visited at least five minutes before he was found hanging). In one case the deceased had been visited 45 minutes before he was found, and in another, he had been taken out for photographs one hour and twenty-five minutes before he was found dead. The time between visiting and finding the deceased in the other three cases were three hours and fifteen minutes, seven hours and forty-five minutes, and twelve hours and thirty minutes.

These cases illustrate that the practice of police visiting people in custody is inconsistent, and that it often appears as though suspects, or people in custody are left alone for long periods during their detention. In two of the last mentioned cases, the lack of visits occurred during the night shifts, and in one case – during the day.

However, where regular visits did occur, it also indicates that these visits may not be sufficient to prevent a person from committing or attempting suicide. In the researcher's later visits to police stations, it was mentioned by police officials that detainees know that the cells are usually visited on the hour, and if they wish to commit an unlawful act or attempt suicide, then they do this once the visit has already taken place.

In one of the 'non-hanging' cases, members of the SAPS arrested 'E', who was a member of a private security firm, at his workplace. The arresting officers had taken his service firearm away from him. He excused himself to go to the bathroom, where he shot himself with his private firearm.

In the other case, 'G' was a reservist police officer who was arrested by a police officer from the anti-corruption unit. He was being conveyed to the police station in a police vehicle. The deceased was sitting in the back of the vehicle, while the anti-corruption unit member and the complainant (a civilian) were seated in the front. The deceased was not handcuffed or restrained in any way. While the vehicle was travelling, the deceased apparently opened the car window and jumped out of the moving vehicle. He sustained injuries. He was put back into the police vehicle, restrained – because he was now thrashing about – and taken to hospital where he later died of his injuries. It is not clear whether the deceased intended to kill himself or not.

Again, in both cases, it appears that the police had failed to take precautionary measures. In E's case, it appears that the police did not search the deceased for his personal firearm, nor did they escort him to the toilet. In G's case, no police officer sat with the deceased in the back seat of the car, and no handcuffs or other restraining measures were used.

Deliberate Self Harm (Pre-custody)

Information concerning the circumstances of the one case recorded in this category were extremely vague owing to the paucity of investigation of the case. The ICD folder consisted solely of the notification report by the police to the ICD. As related by the police, two members of the dog unit were driving in Soweto returning a stolen vehicle. They saw two 'suspicious' males walking ahead, and stopped them to question them. One of the males walked away, pulled out a firearm, and shot himself in the head. Apparently he had previously been arrested and charged on a criminal case. Despite the odd circumstances of this case, the investigation had been closed by the ICD with a finding of 'no police involvement'.

Substance Abuse (Alcohol)

The one deceased in this category died two days after arrest in his cell. The post-mortem report indicated that he died of a medical condition consistent with alcohol abuse. There is no indication in the report, or the docket, of whether the deceased had either been drinking in custody, or had been suffering from withdrawal symptoms.

Substance Abuse (Drugs)

The deceased died from drug related symptoms. In the case of 'O' the deceased was arrested for selling drugs and notes about the post-mortem indicates that he died of drug overdose symptoms. After his arrest, the police took him to his place of residence to point out drugs. At this stage he could not walk and asked for water. He was then taken to the police station to be booked in, and then taken to hospital. The time of death was unrecorded.

It appears that the drug abuse had occurred before being taken into police custody.

Deceased's Medical Condition

Table 5: Deaths in custody arising from the deceased's medical condition – 1998
Deceased's Medical Condition:
no other person's actions associated with the death
Number of Deaths
heart problems1
Heart Problems

One deceased died of a heart attack in the dock whilst giving evidence in the Magistrate's court. He was certified dead when the paramedics arrived to treat him. Since there was little additional information in the docket, it is not known whether the deceased was in police custody, or had been released on bail or warning at the time. Also there was no information about whether he had been receiving or taking medication.


Three cases fit into this category. In two of the cases ('S' and 'T'), the deceased had been prisoners at a prison, and under the authority of the Department of Correctional Services. Despite the ICD allocation of a category of 'death in custody' they were not in police custody at the time of their deaths. One of the deceased had died of bronchopneumonia, and the other had died 'after he was coughing badly'. Although an investigation into the circumstances of their custody, or of any medical treatment received or not received, may have been valuable, this is not the responsibility of the ICD. The newly appointed Independent Prison Inspector has the obligation in terms of the Correctional Services Act, No. 111 of 1998, to investigate cases of deaths of prisoners or those in the custody of the Department of Correctional Services.

The third case in this category is another instance where the deceased had been taken into custody as he was 'mentally disturbed'. He was taken into custody at 0h55 and found dead on the floor of his cell at 07h00 that same morning. No information is contained in the file concerning his health status at the time he was taken into custody. Nor is there information in the file indicating whether or not the police had conducted regular checks on the deceased. The post mortem report indicated that he died of bronchopneumonia and cardiac failure.


One person who had been a prisoner at a prison died of a seizure. He was taken from the prison to the hospital, ICU section, where he died. He was not in the custody of the police at the time.

In the second case, the deceased was arrested for a drug related offence. He tested positive for cannabis. After being taken to court the day after his arrest he began to have seizures and he was taken to hospital where the neurosurgeon noted that he was suffering from convulsions and intra-cerebral haemorrhage. It was noted that he was epileptic. His condition was deemed to be 'inoperable'. He was placed on medication. He had convulsions requiring high care, and it was noted that he was epileptic. He died in hospital after three days.


In three of the five cases under this subsection, no known cause of death is recorded in the ICD dockets. In the case of 'W', he was admitted to the police station as a mental patient. He was also acting disturbed, shouting and punching the cell wall the whole night – reportedly he did not sleep until morning. In the morning he was found dead on the floor of the cell without a blanket. Unfortunately, there is no post mortem report. The police records indicate that 'he was inspected but no injuries were found'. In the other two cases, both deceased were arrested and detained in their cells, and were later found dead. There were no post mortem reports for either of these cases. In one of these cases, the deceased was held in a cell with other people who alerted the police to the fact that he was dead.

The police arrested 'ZW' after some member of the public had complained of a 'drunk' person in the vicinity. The cell register records that ZW was in good condition at 05h00 when he was arrested, although later it was reported that he could not speak, and he had been lying down when the police arrested him. He was found dead in the cells at 09h15 that same morning. Notes about the post mortem indicated that there were fractures on both of his upper arms, and on the side of his head. There were also indications of bleeding on the brain. The notes say that his injuries could have been the result of a fall. The notes go on to say that the symptoms could have been mistaken for drunkenness. Statements in the file state that the deceased was visited several times. The police officer thought he was sleeping and 'left him to sleep some more'. In a later visit it was recorded that the deceased reacted to the visitor, but did not stand up. On a subsequent occasion, the police officer went to visit him and he was 'sleeping on his back'. When he tried to turn the man over, he realised that he was cold – he had died. There was no record of any further investigation into the cause of the man's injuries.

In the last case of this category, 'Z' was arrested and detained by the police. On that same day he was taken to hospital where it was recorded that he died of 'full blown AIDS'.

Summary of Deaths as a Result of Medical Condition

Although there are obviously occasions when people die suddenly and without warning, in most cases there is some indication that a person is not well. If he is sufficiently fit, mentally and physically, the detainee should be able to advise the police of his situation, and obtain medical treatment. It appears that only in the case of Z, who died of AIDS, did the police obtain immediate treatment for him, although it seems he was in an advanced stage of his illness. In all the other cases, there is no indication that any mention was made of the deceased's state of health at the time of his arrest, or subsequently.

It is also disconcerting that the police may mistake a prisoner's apparent drunkenness, or the fact that he is acting 'mad' for a real problem with his health, or a head injury, such as in the case of ZW. Of all the cases scrutinised, five were men who had been detained by the police, not for having committed any crime, but for behaviour apparently connected to being mentally disturbed or drunk.

It is difficult for police officers to identify a head injury or fracture if there are no visible exterior signs. However, 'head injuries may also be indicated by the warning signs … for instance, being insensible, needing to be carried to the cell, and snoring heavily may all indicate problems'.15 Some of these symptoms were present in the cases examined above. Frequent vomiting may also be a sign of injury. In Leigh's UK study several detainees died of inhaling their own vomit. It was therefore recommended that detainees who are vomiting should receive even closer monitoring.

When the ICD is investigating these cases it should pay close attention to whatever warning signs that had been present when the person was arrested or subsequently, and the police's reaction to the signals.

Another Persons Actions May Have Been Associated With The Death

The deaths assigned to this category were instances where the deceased had been injured by another person, either prior to being taken into custody, or while in police custody. It was here that the issue of how to define 'police custody' had the most relevance.

Table 6: Reflects deaths in custody where another person's actions may have been associated with the death – 1998
Another Person's Actions May Have Been Associated:
Actions of another person may have been associated with the death
Number of Deaths
Police actions may have been associated4
Doctor/medical action0
(one instance were paramedics declared person fit)
Other person's action while in custody1
Other person's action (pre-custody)16
Police Actions May Have Been Associated

In all four of the cases under scrutiny, the deceased had died as a result of gunshot wounds fired by the police. In all of the cases, the deceased was ostensibly attempting to escape from police custody after having been taken into custody.

In the case of 'ZA', the deceased and others had been arrested by the police and had been taken outside the building where they were arrested. The deceased and another suspect tried to run away. After shouting a warning, and firing a warning shot, the police officer fired at the deceased, shooting him in the head. He died before the paramedics arrived.

The case of 'ZB' was similar. He had already been booked into a cell, and the police were taking him out on further investigation. Apparently his handcuffs had been removed just prior to the incident so that he could sign a statement. Both police officers got out of the police vehicle. It is reported that the deceased took this opportunity to run away. Despite shouting a warning, and firing a warning shot, the deceased continued running. He was also shot in the head. He died in hospital.

The police had arrested the deceased, 'ZD' earlier in the evening, together with other suspects. The police were doing follow-up investigation, when the deceased allegedly jumped out of the police vehicle and ran away. He did not heed verbal warning or warning shots. The police member then fired two shots at the deceased which fatally injured him. He sustained chest and abdominal gunshot injuries.

In the last case 'ZC' was already being held in the police cells. While one policeman was counting prisoners in the cells, one of the prisoners grabbed his firearm and shot at the policeman, wounding him in the head. Another policeman heard the commotion and came to his aid. The prisoner shot at him, injuring him in the leg. The police officer returned fire fatally injuring the deceased and wounding two other prisoners. All the injured parties were taken to the hospital for treatment. The ICD found that the police action was justified in this case, although some questions remain unanswered. It seems that contrary to police policy that the police official was counting the prisoners alone, while he was carrying a firearm making himself vulnerable to attack.16

The alien investigation unit arrested one of the suspects, presumably on suspicion of being an illegal alien. After he was shot, it was also alleged that he had attempted to bribe a police officer. In the other cases, the suspects had been arrested for robbery; armed robbery; and hi-jacking – all serious or schedule one offences.


In the cases of 'ZA', 'ZB', and 'ZD', the suspects had all attempted to escape while under direct police control. In none of these cases were the deceased under any form of restraining measures to ensure that they did not escape. Had they been restrained, their attempted escape would have been more difficult and the extreme measures taken by the police may have been prevented. Police policy and practice needs to be re-examined to ensure that retraining measures are used, both as a preventative measure against escape and against deaths occurring in custody.

In the case of 'ZC', the deceased was able to grab the police officer's firearm while he was in the cells. A preventative measure that could have been taken in this case was not to carry the firearm while inspecting the police cells. This should be a clear policy decision applicable to all security officials while visiting prisoners and detainees in police and prison cells.

Doctors'/Medic's Actions

Although not reflected in the table under this category, there is one case where the actions of the paramedics may have contributed to the death of the deceased. In the case of 'ZN', members of the public had assaulted the deceased after being suspected of burglary. He had been hit on the back of the head with an axe. He was injured and seen by the paramedics who declared him fit. He was then taken to the police station to be booked in, but he collapsed. The paramedics arrived again, and stabilised him and took him to the hospital where he died. The time between the original injuries, the first time that the paramedics saw him, and the time that he finally collapsed is not recorded. However, in an injury of this nature, the question arises as to whether the paramedics gave him a thorough examination, and whether he should instead have been taken straight to the hospital.

Other Person's Actions While in Custody

The facts in this case were not very clear, as the investigation was still pending. It appears that 'ZF' was arrested by the police and taken to point out another suspect. It is alleged that the person he pointed out became angry with the deceased, and in the presence of the police, assaulted him. ZF was later taken to the police cells and booked-in. He did not complain of any injuries, but a scratch was noted under his eye, and he was detained with twelve other inmates. Several hours later he was taken out for further investigation. He died in the presence of the investigating officer shortly thereafter. The post-mortem report noted that the cause of deaths was not ascertained, although several abrasions were noted on the body, and there were haemorrhages in the heart and lung. There were also signs of anoxia.

Other Person's Actions – Pre-custody

This group of cases constituted the greatest category of deaths. In most instances, the deceased was assaulted, or sustained injuries at the hands of one or more members of the community, either in the course of committing a crime, or on suspicion of having committed a crime at some other time. Many of the latter cases bore the marks of spontaneous vigilante action by a group in the community. In all but one case, the assault occurred long before the police arrived, and the deceased was already fatally injured. It was left to the police to arrest the suspect and take him to hospital for treatment. In many cases, the police probably cannot be held responsible for the deaths, or for contributing to the death of these people. However, this category of deaths deserves its own investigation. Important issues to pursue in these cases would be to determine whether the police did everything they could to prevent the death from occurring, or to obtain medical treatment for the deceased at the earliest opportunity.

An investigation on the time taken for the police to arrive at the scene after receiving the initial report would be a crucial issue. In a study conducted by the CSVR into police attitudes towards vigilantism, the police indicated that there might be a number of reasons why they may fail to intervene in a situation involving vigilante action. They said that they were several risks in becoming involved in that type of mob action, either a risk to their personal safety, or a risk of becoming associated with the action itself. One officer said that when the police 'receive a report that a group of people are beating someone, they will frequently simply ignore the report and wait until the incident appeared to be over. At this point they might go to the scene and ensure that the beaten person is taken to hospital'.17

Yet in the majority of cases the ICD investigators did not follow up these questions. Their investigation seemed only to have the primary objective of ascertaining whether the injuries were sustained at the hands of the police or the community. Once this was established, the investigation was closed.

In some of the cases the police stated that a charge had been laid against a member or members of the community. Since this was not followed up by the ICD, it is not clear whether the police pursued these investigations.

In one case the deceased was attacked by members of the community on an allegation of rape and murder. Several police officers arrived on the scene, but were told to keep away by the group. In the police's presence, members of the group poured paraffin over the man and set him alight. The police called for reinforcements and managed to rescue the man. An ambulance arrived which took him to hospital. He died of internal injuries on the way to hospital.

Gender And Racial Breakdown Based on Gauteng Docket Analysis For 1998

No women were found in the sample of dockets scrutinized. In most cases the race of the deceased was not stipulated. The researchers frequently made judgements based on the name of the deceased, and consequently the classifications should be treated with a high degree of caution. In several instances the age of the deceased was also not known.

Table 7: Indicates the gender, age and race group of the people who died in custody during 1998
AgeRace UnknownAfrican MaleIndian MaleColoured MaleWhite MaleFemalesTotal
unknown age2171 1 21
19 and younger 4    4
20-25 years 5    5
26-30 years 7    7
31-35 4 1  5
36-55 years 2  3 5

The majority of people who died were African. This may be related to the demographics of the country, and also that the majority of people held in police custody are African. Although most of the ages were unrecorded, most of deaths of those whose ages were known to the researchers occurred in the 20 to 30 age group. This can be contrasted with the males in the Coloured and White groups who died most frequently in the older age groups – 31- 35 in respect of Coloured, and 36 – 55 in respect of whites.

Of the people who died as a 'result of deceased's own action- deliberate self-harm' most of the deceased were in the 24 – 29 age group. One of the deceased was 30 years of age, and another was 46 years. In the sub-category of 'substance abuse' however, one deceased was 31 years old. The age of the other deceased in this category is unknown.

Of the people that died 'as a result of deceased's medical condition' all but two were older than 30 years of age. Their ages were 19, 28, 32, 34, 42, 45, 48, 50, 51 and 53. The age of one of the deceased was unknown.

In respect of the people who died where 'another person's actions may have been associated' with their deaths, all of those whose ages were known to us, were of or under the age of 35 years old. However, the ages of 62% of the people who died as a result of 'another person's action (pre-custody)' are unknown.

Breakdown According to Policing Area

In Table 8 below, the number of deaths are analysed according to policing area, and these figures compared with the numbers of prisoners held in police cells in each area. However, these calculations are based on the Gauteng dockets scrutinized by the researchers. Whereas only 47 dockets relating to deaths in custody were scrutinized by CSVR, the ICD recorded 60 deaths in police custody for the relevant period. If all the dockets were included in this table, there may have been significant differences in particular areas.

Table 8: Comparison between deaths in custody in each policing area, and numbers of people held in police cells – 1998
Gauteng policing areaOther persons action pre-custodyAll other deathsNumber of prisoners held in police cells per policing area18 for 1998Deaths per 100,000 people taken into custody
Vaal Rand4229.00620.68
Johannesburg Prison/Magistrates Court 4  
West Rand0161,6101.62
East Rand4364,87710.78
North Rand3542,45318.84

These figures distinguish between deaths that occurred as a result of another person's actions before being taken into police custody, and other deaths in police custody. The rate clearly varies quite dramatically from area to area.

The overall rate of deaths in police custody is calculated at 12.31 per 100 000 prisoners held in the police cells. Had this report included all the deaths recorded by the ICD, the overall rate of deaths per 100,000 would have increased to 15.72. This can be unfavourably compared with the rates for deaths in custody in England and Wales. A study conducted by the Home Office in the United Kingdom finds that between 1990 and 1997, there were an estimated 3.2 deaths per 100 000 arrests for notifiable offences. The study cautions that if the number of deaths were calculated as a proportion of the true number of people passing through police custody, the figure would have been far lower.19

Deaths in police custody occur at the highest rate in the Vaal Rand, followed by the North Rand, and Soweto. In the Vaal Rand, most of the deaths are attributed to the actions of other persons before being taken into custody. In Soweto and North Rand, most of the deaths occur in police custody. Stations in these areas need to examine their particular management practices in relation to people in custody. Despite the high number of people arrested and taken into custody in the Johannesburg area, the rate at which deaths occur is just below the overall rate (12.31 deaths per 100,000 of the population). However, most of the deaths occurred in police custody, rather than being attributable to people's actions outside of police custody. Again this indicates a need to look at particular management practices regarding people held in police custody. A similar situation exists in respect of Pretoria. According to the figures, the West Rand indicates the lowest rate of deaths, as only one death was recorded as having occurred in this area. This is despite the large number of people arrested and taken into custody.

If this analysis was conducted including all the cases of deaths in police custody reported to the ICD, the results may be substantially different.

Section Two: The Management of People in Police Custody

International Principles Governing Detention of Persons

It is only recently, since the new democratic order in this country, that international principles have had significant influence over the arrest and detention of people in South Africa. South Africa has adopted the Universal Declaration of Human Rights. This instrument recognises the inherent dignity and the inalienable rights of all human beings. It specifically guarantees the right to life, liberty and security of persons (Article 3), and prohibits cruel, inhuman and degrading treatment or punishment (Article 5). In 10 December 1998, the South African government ratified the United Nations Convention Against Torture and Other Cruel Inhuman and Degrading Treatment or Punishment (1984). This convention outlaws torture. It obliges states to take effective legislative, administrative, judicial or other measures to prevent acts of torture, and ensures that acts of torture become offences under the criminal law. Although primarily concerned with torture, the CAT also says that each party shall undertake to prevent other acts of cruel, inhuman or degrading treatment or punishment which do not amount to torture (Article 16). South Africa also ratified the International Covenant on Civil and Political Rights (ICCPR, 1966) in 1998. In addition to prohibiting cruel, inhuman or degrading treatment or punishment and torture, the ICCPR also requires that 'all persons deprived of their liberty shall be treated with humanity and with respect for the inherent dignity of the human person' (Article 10).

South Africa ratified the African Charter on Human and People's Rights (1981) in 1996. Article 5 enshrines the right to be treated with respect of the 'dignity inherent in every human being', and prohibits 'all forms of exploitation and degradation of man particularly slavery, slave trade, torture, cruel, inhuman or degrading punishment and treatment'.

In addition to these major instruments, there are various international instruments that establish international principles for the detention of persons. The most widely known and applied is the United Nations Standard Minimum Rules for the Treatment of Prisoners (1957). Although South Africa is not a signatory, these standards are influential on our law, and serve as the basis for the new Correctional Services Act of 1998. The rules cover the general management of institutions, and are applicable to all categories of prisoners, criminal, civil, untried or convicted, including prisoners subject to 'security measures'. Another important United Nations instrument is the Body of Principles for the Protection of All Persons under Any Form of Detention or Imprisonment (1998). The Code of Conduct for Law Enforcement Officials (1979) and the Basic Principles on the Use of Force and Firearms by Law Enforcement Officials (1990) are also relevant to the management of people held in police custody.

South African Legislation and Policy

The South African Constitution of 1996 also contains provisions prohibiting torture and cruel, inhuman or degrading treatment or punishment, and it enshrines the right to human dignity. The Bill of Rights, in Section 35, deals specifically with the rights of arrested, detained and accused persons, and in particular accords them their basic human rights. This provides the police with the basis for how to treat people in their custody.

The South African Police Service Act is largely silent on the role of the SAPS in relation to the detention of people. This is dealt with by the Criminal Procedure Act, No. 51 of 1997. Section 50 of the Act deals with procedure after arrest, and mandates that 'any person who is arrested with or without a warrant for allegedly committing an offence, or for any other reason, shall as soon as possible be brought to a police station' or any other place mentioned in a warrant of arrest (S 50(1)). The section also provides that if the person is not released earlier, he/she should be brought before a lower court within 48 hours of arrest. The court can make an order that the person be further detained (S 50(5)). The Act does not outline the conditions under which a person should be held in detention, nor does it outline the various responsibilities of the South African Police Service.

The South African Police Service Act does not deal with the arrest and detention of accused persons. It is the Police Standing Orders that govern police activities in relation to persons in custody.20 The Standing Orders contain much detail on the administration of persons in custody, and regulate such things as their admission, recording of information on admission, the bedding in cells, instruments of restraints which may be used, visiting of police cells by members, meals, prisoner's work, conveyance of prisoners, etc. The Standing Orders date from 1991, amended in 1992, and in many instances fail to take account of international principles, or even of the Constitution.

In order to bring the policing practice up to date with recent developments, a Policy on the Prevention of Torture and The Treatment of Persons in Custody of the South African Police Service has been developed by the SAPS. This document sets out instructions, which station commissioners, and other commanders are required to enforce in the areas under their authority. This Policy document attempts to give effect to Constitutional principles, as well as to provide some detailed instructions on the management of persons in custody within this framework.

The Policy on Prevention of Torture defines torture in exceptionally broad terms and states that:

Torture may include, but is not limited to, any cruel, inhuman or degrading treatment or punishment, as referred to in section 12(1)(e) of the Constitution of the Republic of South Africa … or any act by which severe pain suffering or humiliation, whether physical or mental, is intentionally inflicted on a person for the purposes of obtaining from him or her or a third person information or a confession, punishing him or her for an act he or she committed or is suspected of having committed, or intimidating him or her or a third person, when such pain, suffering or humiliation is inflicted by or at the instigation of or with the consent or acquiescence of a member or any other person acting under the authority or protection of the (South African Police Service).

In terms of the definition, 'cruel, inhuman or degrading treatment or punishment' is also defined as torture, and not as a separate type of treatment, whereas the international instruments distinguish between torture and cruel, inhuman or degrading treatment or punishment. This places the police under the same obligation to prevent cruel, inhuman or degrading treatment or punishment, as it does to refrain from and to prevent torture.

The Policy on the Prevention of Torture was practically implemented towards the end of 1998 to the extent that the guidelines were supposed to be adhered to at the police stations. The policy required certain notification forms to be completed in respect of persons in custody. However, since it also conflicts with the Standing Orders of 1991 and 1992, these Standing Orders are in the process of being re-drafted, and once approved, it is expected that they and the Policy will be fully effective from 1 July 1999.

In essence, incarceration of arrested people in police cells serves to prevent suspects from obstructing the course of the investigation, either by fleeing or interfering with witnesses, and to facilitate investigations by keeping suspects available to the police. At the same time the custodians must ensure that the voluntary nature of suspects' statements is not jeopardised.21 Thus the policies, laws and practices should be designed to ensure that suspects are protected against potential abuse of power by the police.

While the police have received much criticism for treatment of detainees, torture and long periods of detention without trial, there has been relatively little focus by human rights activists and others on the conditions of police custodial facilities, as opposed to conditions in prisons. This report aims to shed some light on the treatment of 'ordinary people subject to police detention'.

In the following sections this report looks at the conditions in police cells, and the management practices in light of the international and Constitutional principles, and domestic laws and policies.

Visits to Police Stations in Gauteng

During April and May 1999, ten police stations were visited in Gauteng.22 The police stations were randomly selected from a list of police stations provided by the Gauteng Department of Public Safety and Security. Sixteen people were interviewed at these police stations. Five of these people were station commissioners, and one was an acting station commissioner. We also interviewed five client service centre commanders of the ranks of Sergeant, Captain and Inspector. Three cell commanders, with the rank of captain and inspector, were also interviewed. In addition, we interviewed a head of the pro-active unit (Superintendent) at one station, and the Commander of the Guard Unit (Inspector) at another.

The police cells were visited at each police station.

Supervision of police cells

Depending on the size of the police station, the number of cells, and the number of prisoners accommodated, the police had different structures for the management and oversight of prisoners in the cells. The Station Commander bears final responsibility for the cells, and usually did visit the cells at least on a weekly basis. At the smaller stations, the Community Service Centre Commander, was responsible for management of the cells. At a larger station, the cells were the sole responsibility of an inspector. In another unit, the head of the pro-active unit was the person responsible.

At most of the police stations, there was no specific allocation of police officials to manage the cells. In most cases, the task of hourly visits to the cells, maintenance of cells registers and attendance to complaints, etc., fell to the police officers on duty at the Community Service Centre at any particular time. Perhaps one or two individuals would be given that particular responsibility on any shift. However, at three of the stations visited, the cell complex included an office which was staffed by two or three members during the day, and two at night. One police station had a part time cell guard who took care of the prisoners, as well as attending to court duties.

Training and Qualifications

Art. 1 of the Code of Conduct for Law Enforcement Officials says that

Law enforcement officials shall at all times fulfil the duty imposed upon them by law, by serving the community and by protecting all persons against illegal acts, consistent with the high degree of responsibility required by the profession.

Article 2 goes on to say that, in the performance of their duty, law enforcement officials shall respect and protect human dignity and uphold the human rights of all persons. In order to be able to implement human rights principles, and manage persons in custody in an effective and humane manner, it is essential that staff members receive appropriate training.

The more senior policing officers we interviewed had received basic policing training and some managerial training. Only two of them had received any form of human rights training. They had also received training on the rules and regulations applicable to the police. The researchers asked all the police interviewed what training they had received around the treatment or management of people in police custody. Other than what is provided in the basic training, and through departmental orders, none of the police had received any training in this sphere. One person said, 'Nowhere in the country are prison guards trained to treat prisoners, other than what the law says'. In one police station, the cell commander said that they usually held fortnightly briefing sessions with the staff where they discussed new departmental orders and guidelines. It was stated that most of the officers learnt about the treatment of prisoners on the job. Only two of the officers spoken to mentioned the SAPS Prevention of Torture Policy.

Many of the interviewees did mention that training would be a priority and could help in the better management of prisoners.

Human Dignity

In terms of international instruments and our own Constitution,23 prisoners are required to be kept in conditions which are consonant with human dignity. The Body of Principles and the Standard Minimum Rules provide detailed guidelines on what conditions should prevail where people are held in custody. The SAPS Prevention of Torture Policy also contains similar provisions.

Prisoners are typically held in police stations for very short periods, usually not longer than 48 hours, after which, if they are still held in custody, they are sent to be detained in prisons. However, situations where prisoners are held for periods of up to two weeks are not unusual.

In addition to holding awaiting trial or newly arrested prisoners, all the police stations also hold sentenced prisoners who are due for release. These prisoners – termed 'hard labour's' – are usually required to perform work around the police station. They can be held in police cells for a month or longer.

Cell capacity

The SAPS Prevention of Torture Policy states that the

… number of persons who are accommodated in a police cell shall be reasonable in relation to the size thereof. If this is not possible, the station commissioner shall accommodate all additional persons in suitable alternative accommodation, such as an office or in the police cells of a neighbouring station. (Paragraph 6(1)(A)(i))

The size and capacity of the police cells varied at each police station. Stations usually had a mixture of communal and single cells. The communal cells differed in size, having been made to accommodate 20 or 10 people. A single cell typically measured approximately 3 metres by 3 metres. In other stations, a cell measured 4 metres by 4 metres. Cells were usually built around a communal courtyard – sometimes three to five cells to a courtyard. The single cells were situated around their own courtyard. In all of the prisons, the courtyards were covered by wire mesh for security reasons. A shower and basin were situated in the courtyard.

Most of the police cells were operating under capacity, with fewer prisoners held in each cell than they could accommodate. The two exceptions were Springs and Kathlehong police stations. Springs also had the oldest holding cell complex visited by researchers, and the cells appeared to be cramped. One small cell, of about 3 by 3 meters, was accommodating 4 child24 prisoners. These children had also been held in the cells for up to two months, due to a lack of alternative accommodation. The courtyards adjacent to the single cells were smaller than usual. Many prisoners also populated a complex of three cells leading onto one courtyard. One of the cells in this complex was insecure, as the window could easily be broken and some prisoners had previously escaped from it, hence it could not be used for sleeping accommodation. We were advised that this police station frequently experienced a problem of overcrowding. Although their daily capacity was at around 68 prisoners, over weekends they sometimes have to hold more than 100 prisoners. Where possible, the station referred the excess prisoners to other police stations. However this station, as well as others in our research sample, was accommodating prisoners arrested at other police stations which had no detention facilities. This exacerbated their accommodation problems.

At Kathlehong police station, the researchers were informed that they sometimes had to hold up to 40 prisoners in cells designed for 10 people. The researchers were advised that this station was holding between 200 – 300 people for ordinary crimes, and 150 – 200 illegal immigrants each month. This station had only four communal cells, of which one was for women.

One of the ICD investigators interviewed said that overcrowding was a severe problem in police cells. He related the story of one prisoner who had attempted suicide. The prisoner said that he had wanted to commit suicide because the cells were so overcrowded, and the conditions so shocking that he felt ashamed to be detained in them. In circumstances of overcrowding, combined with poor facilities, conditions of detention could be said to be cruel, inhuman and degrading.

All of the police stations accommodated women and people held in respect of illegal immigrations separately. The sections for these foreigners were sometimes more crowded than the other sections, although these prisoners tended to get moved out of the police station to appropriate institutions (such as Lindela) more frequently.

Few of the police stations accommodated child prisoners, stating that they were either detained in prisons, or in special places of safety for children awaiting trial. The Constitution states that children should not be detained except as a measure of last resort, and that when they are, this should be for the shortest period possible. Section 28(g) goes on to say that child detainees or prisoners should be detained separately from persons over the age of 18 years of age, and in the police stations visited this seemed to be the case. However, Section 28(g)(ii) also says that such prisoners should be 'treated in a manner, and kept in conditions that take account of the child's age'. Despite this provision, the cells holding children were no different from those holding adults, and in the one case, they could even be said to have been more cramped and dark than the cells accommodating adults at the same station.

In accordance with international principles sentenced prisoners were always held in a separate cell to un-sentenced prisoners. There were usually fewer sentenced prisoners held at each police station and accordingly their conditions were less cramped. Since the prisoners were required to work at the police station, they were not locked up all day as the other prisoners were. These prisoners had made an effort to keep their sections neat and clean. At a few of the stations, they were sleeping on proper beds.

Cell conditions

Paragraph 6(1)(a) of the SAPS Prevention of Torture Policy states that

  • all police cells should have adequate light and ventilation,
  • cells should be equipped with reasonable means of rest such as a chair or bench,
  • if a person is kept overnight in custody, he/she should be provided with a mattress and blankets of reasonable standard,
  • a person in custody must be allowed to use toilet facilities and be offered adequate washing facilities,
  • open air exercise must be offered daily where possible.

All the cells had windows that led to the outside. In all but one case these were covered with either wire mesh, or bars. In some cases, the wire mesh significantly diminished the amount of fresh air and natural light that was allowed into the cells. At one police station the windows did not open at all to allow the passage of fresh air due to the security fixtures on the window. We were advised that fresh air could enter through the door of the cell. However, it was said that this door was kept closed at night.

In one prison, the windows consisted of double-glazing and did not open to allow the passage of fresh air. These cells were instead equipped with air conditioning units which were said to be unreliable.

All of the cells were gloomy, and most had an electric light switched on, even in the middle of the day. In several cells, the lights were not working in all.

The cleanliness of the cells also varied from police station to police station. In several cells, the cells were unswept and papers and other litter blocked the drains. Most of the cell walls were covered in graffiti or burn marks, even though they were washed occasionally. Many of the police officers complained about the conditions of the cells, saying that they needed painting. In some of the stations, it seemed that the battleship-grey walls had not been painted since the police cells were built, and in the courtyard of one station, the paint was hanging in strips from the wall. Many of the officials said they needed money to upgrade the cells and particularly to paint the cells. They thought it was important to make the cells better to live in. However, the cells were clean in many of the police stations visited, as were the surrounding corridors and areas.

Most of the cells had toilet facilities and basins in the cells, as well as in the courtyard. Except at one police station where a cell had been allocated specifically for female prisoners, the toilets were not screened from the main portion of the cell, and could not afford any privacy. These were of a standard design, which we are advised was fairly new in the cells. One police officer complained that the 'new design' was not very efficient and said they were prone to difficulties and frequent blockages. Other officers complained that more toilets were needed in the larger communal cells. A showerhead was situated in the courtyard of the cells.

Some of the police officials interviewed complained of maintenance problems in the police cells, and stated that even though they had asked for simple maintenance tasks to be performed, there were often long delays before these were carried out. This could be seen, especially in the older cells, where there was a problem of drainage. There was often un-drained water in the washing areas, and in one station a particularly acrid smell – the result of bad plumbing.

Some of the cells had built-in concrete bunks to sleep on, although there were fewer 'bunks' than would be required in the communal cells. All inmates were provided with a felt mattress and grey blankets. One interviewee advised that the blankets were washed on a fortnightly basis. None of the cells provided any form of chair, or other 'means of rest'.

All the cells were built around a courtyard, and we were advised that the cells doors were unlocked so that prisoners could make use of the courtyards for a large portion of the day.

According to ICD investigators, the conditions of Gauteng police cells are much better than those in other provinces. However, one senior ICD official said. 'Conditions in police cells could be distressing. I don't think that they present a good environment for people to accept that they have been detained. This could be one factor contributing to their suicide'.

In many instances, the conditions of the cells was not inconformity with the principle of respect for the human dignity of a person, and in several cases, could also be said to be inhuman or degrading.


One of the major functions of the police in detaining people is to ensure they are held in safe custody and that they do not escape from police custody. The high number of escapes (a national increase from 3 595 in 1996 to 3 818 in 1998 ) is a source of some concern to the police. Three factors could contribute towards the incidence of escapes from custody: negligence on the part of police members; corruption or collusion in escapes; and inadequate or insecure holding facilities.

Most of the cells had bars on the cells doors, and mesh or grids on the windows and metal grids over the courtyards. In special security cells, such as in Sophiatown, the security was even tighter. The cells doors were locked with an Allen Key, and each cell had two outer doors. The security section was further separated from the rest of the section by another secure door. Despite this high security, these cells have not been in use since approximately 1995. Krugersdorp was the only police station visited by researchers that had taken extraordinary security measures. A project supported by local businesses had resulted in each cell being fitted with electronic security alarms. There were infrared sensors that detect human movement and wires on the roof and over each of the cell windows. When inspecting the cells, the police officers carried a panic button that would set off an alarm when pushed, as well as indicating the location of the police officer on a control board in the cells offices.26 If the security was breached elsewhere this would also set off an alarm linked to the control board.

In other police stations, the members complained that it would be possible for any prisoner to escape if they tried hard enough. At every station the researchers were told about prisoners who had either tried, or who had succeeded, in escaping. However, it did seem that the problem was diminishing in the areas visited. The police also complained that when the security had been compromised, for example where a window or lock was broken, it often took a very long time for these to be repaired. The result was that either, prisoners could not be held in particular sections of the cells, or they were held in not very secure conditions.

It should be noted that in the cases observed by CSVR, only one person had allegedly been shot while attempting to escape from the cells. In this instance, the attempted escape had occurred in the presence of police officers.

Another important objective of the police is to ensure the safety of the prisoners whilst they are in custody. A relevant issue here is whether the police are able to respond promptly to a situation where a prisoner is in danger or requires help.

No cells were equipped with alarms that a prisoner could use to alert the police. They were reliant on shouting for help. In many stations, the cells were situated at some physical distance to the main part of the police station, and police would not be able to hear any call for help. Even where the cell guards did have an office adjacent to the actual cells, this was often some distance away from many of the cells, and was often separated by a closed door. In many cases it was likely that the police would only become aware of a problem during one of their routine visits.

Reading material

Section 35(1) of the Constitution says that every detained person is entitled to be provided with reading material at state expense. This requirement did not seem to be adhered to at any police station, as no prisoner was seen in possession of any reading material.


A person in custody has the right to communicate with and be visited by that person's spouse or partner, next of kin, chosen religious counsellor, and medical practitioner (Section 35(1) of the Constitution). Visits from family members were allowed at each police station, although facilities varied enormously. Only a few of the police stations had a separate visiting room where prisoners and visitor could communicate through glass or a grid. In some places, the visits took place in one room, where they were watched over by a police officer. In several police stations, the visits took place through the gate at the entrance to the cells with the visitor standing outside. Most police stations had limited visiting hours, although exceptions could be made in special circumstances.

One station set aside a room for prisoners to make telephone calls.

Rights of detained person and right to consult with a legal practitioner

Section 35(2) of the Constitution says that everyone who is detained has the right to:

  • be informed of the reason for being detained,
  • to choose, and to consult with a legal practitioner, and to be informed of this right,
  • to have a legal practitioner assigned by the state and at state expense if substantial injustice would otherwise result, and
  • to challenge the lawfulness of his/her detention.

The SAPS Policy says that a person taken into custody shall be informed of these rights by the member who takes him or her into custody, in a language that he/she understands. In order to ensure that this process occurs, the SAPS Policy states that the community service centre commander must give that person a written notice setting out his or her rights on arrival at the police cells. At one police station we were shown a copy of such a form, and the place in the cell register where it is recorded that the person in custody has been informed of his rights. It was noted however, that the appropriate column in that cell register had not been ticked in the majority of cases.

In several police stations, but not all, there were signs at the intake sections, and at other places in the cells which detailed prisoners' rights. One police station had signs that advised the detainee to contact the Independent Complaints Directorate if they had any complaints against the police.

Legal visits were allowed in the stations visited, although they may take place in slightly different circumstances where no visiting facility was available. For instance, one police station set aside an office in the cellblock for this purpose; others used an office in the main police station.

One officer complained that prisoners demanded an excessive amount of phone calls, or wanted to be able to use their cellphones. He said that only one phone call was allowed, and the use of cellphones was prohibited.

Management of prisoners in need of special care
Prisoners under the influence of drugs/alcohol

The SAPS Prevention of Torture Policy states that any person in custody 'who is under the influence of intoxicating liquor, drugs or medicine, or who is in a state of shock' shall not be questioned by the police (Article 10(6)). No other mention is made in the policy for the protection or care of this category of prisoner. The old SAPS Standing Orders however, do contain provisions. Order No. 361.15. says that, 'if a prisoner is insensible from drink or any other cause, he shall have his throat laid bare and shall have his head raised when lying down'. It also says that he should be visited at least every half hour until he has recovered consciousness. He should be also be roused at each visit unless he is breathing regularly.27

When asked how police stations managed inmates who were drunk or insensible from any other cause, pretty standard answers were given at each station. The researchers were advised that these people would be held separately, and visited every half hour by the cells guards. Once the person sobers up he can be held with the other detainees. At one police station the respondent mentioned that they regularly pick up the same people for being intoxicated, and it often happened that these people were so drunk that they could not walk.

The police officers interviewed said that they did not have problems identifying prisoners with special needs. Usually, they said, a prisoner will advise you if there is a complaint. If a person is drunk, this can be detected by the smell on their breath. Often the investigating officer would alert the station to the prisoner's state of health.

However, an ICD investigator said that the police often disregard information or complaints made by a prisoner who is drunk. He said, 'The police will say he is drunk, or a criminal', and so the complaints don't matter.

People suffering from mental impairment

The SAPS Policy does have special protective guidelines people who appear to be mentally impaired, but only in relation to conveying the person's right to communicate with a legal practitioner. It says that if a person is mentally handicapped or is suffering from a mental disorder, the community services centre commander 'shall, as soon as is practicable, inform an appropriate adult of the grounds for his or her custody and the place of detention', and ask them to visit the detained person at the police station (Paragraph (2)(d)(iii)28 and (iv)). No other special measures are indicated.

The Policy also restates a prisoner's right to communicate and be visited by a medical practitioner of his/her choice. Paragraph (4)(a) says that if a person

… appears to be suffering from physical or mental illness, is injured, does not react to sensory stimulation, displays a lack of awareness of his or her surroundings or otherwise appears to need medical attention, the community services centre commander or any member in control of a person in custody shall immediately call a district surgeon or, in urgent cases, send the person to the nearest provincial hospital available or call the nearest available medical practitioner.

Neither the Policy, nor the existing Standing Orders prescribe guidelines on how to treat a person suffering from mental illness. From our analysis of the dockets, and from respondents we interviewed, the police are often required to deal with people suffering from mental impairment. SAPS members are unskilled at diagnosing the mental state of detainees, and at times appear (from the docket analysis) to mistake mental illness or injury, for drunkenness. This is potentially dangerous, because if left alone, the person might injure himself or others. If the illness is serious, lack of treatment could lead to serious health problems, or death.

Despite the lack of direct guidelines, most police stations had their own system for dealing with mentally ill detainees. One police officer complained that mentally impaired people are often violent and difficult to manage. The primary concern of the police seemed to be to have them removed from the police station to an institution which can cater for their needs. The problem raised however was that it may take a whole day of administrative procedures to get the magistrate to make an order of committal, for the district surgeon to make a notification, and to find a hospital which would take the detainee. As a result of these delays, or due to the person being admitted late in the day or in the evening, these people may be held in custody for a night. At another police station, it was said that if the person is acting strange, they would call the investigating officer to deal with the problem.

Physically injured and sick people

In respect of people who are admitted when they are ill or physically injured, it was said that a SAP 70 form would be completed, and the person would be taken to hospital for treatment. If the person was admitted to hospital, a police officer would have to stand guard over that prisoner. One police officer said that because the State becomes liable for the costs of medical treatment of injured prisoners, in some cases where the suspect is badly injured, they would withdraw charges against that person, until he/she recovers sufficiently to be released. The hospital notifies the SAPS of their impending release. Charges could then be re-instated.

We were advised that if a person becomes ill while in custody, the people in the shift notify the commanders and the SAP 70 form be completed. The district surgeon would be notified, or the prisoner would be taken to a clinic or hospital. Sometimes paramedics are called out to the police station.

The SAPS Prevention of Torture Policy states that a custody register should be kept at each police station wherein every action taken by a member of the SAPS in relation to the person in custody, should be recorded. The community service centre must ensure that a record is made of 'any injury suffered before, during or after arrest', and must record the circumstances in the Occurrence Book (paragraph 8(2)).

An ICD investigator said that it was often a problem at police stations that an illness of a prisoner was not properly recorded, and where one person knew of this information, it was not properly handed over to the next shift.

If it is a tendency that the police do not record the injuries or illnesses sustained by a person who is admitted with injuries, then it is even less likely that police will record injuries that are the result of assault or torture inflicted by members of the SAPS. This raises the question as to how effective the SAPS Prevention of Torture Policy will be in practice.

Suicidal people

Suicide is one of the major causes of deaths in police custody. According to ICD Statistics, 25% of people (52) who died in police custody in 1998 are recorded as having died as a result of suicide. The statistics do not indicate the number of people who may have attempted suicide, but who survived. Two of the police stations that were visited, mentioned cases of attempted suicide. In one instance, the person had successfully been prevented from committing suicide whilst in the cells, but once he was released on bail, he had killed himself.

Respondents were asked how they identified suicidal prisoners and what actions they took to prevent them from committing suicide. One police officer said that it is 'difficult to know when a person is at risk – unless he speaks about his intention'. Another police officer said that people will often threaten that should they be locked up, they will commit suicide, but then do not follow through on the threats. The police tend to only learn that someone is suicidal once that person attempted suicide, or had completed the act. If the prisoner is held with others in the same cell, the cellmates would usually alert the guards, or would try to stop the person from any suicide attempt. Sometimes, it was said, the cell guards may notice a problem when they do their visits.

Ways in which police stations dealt with suicidal people were quite limited, and usually entailed putting the person with others in the communal cells, and searching the detainees and removing items from them which could be used to injure themselves, such as belts, shoe laces, etc. Police also noted that prisoners might use other items such as blankets or towels. This is confirmed as four people in our sample had used items that they were allowed to keep with them in the cells, such as a torn blanket, trousers and elastic from trousers, and bandages, which they used as ligatures to hang themselves. Police members also said that if there were a known risk, they would visit the detainee in the cells every half an hour.

One problem noted by one police officer interviewed was that 'the period for which a person is detained, and our budget, does not allow us to deal with (suicidal people) in any more extensive way'. An officer at another station said that they had police or civilian staff at the station who were trained on victim empowerment, and they could ask these people to speak to the suicidal person. Another option is to elicit the services of a social worker. However, at this station, the threat of suicide had not yet become a problem, so it could not be said whether this method would be effective. Only one police station said that some of their officers had been trained on dealing with problems of suicide, and all cases of potential suicide are referred to them.

The Prevention of Torture Policy provides that a custody register shall be kept at each police station. Any injury suffered by the detainee before, during or after arrest, and the circumstances thereof, should be recorded in the Occurrence Book. Any other medical information such as when a person in custody underwent a medical examination should also be recorded in the Occurrence Book and cross-referenced to the Custody Register (Paragraph 8).

However, there seemed to be no procedure for screening detainees or for making an assessment on admittance to ascertain suicide or health risks. Although it was mentioned that 'if the cells guards noticed something wrong' they would make an appropriate intervention, there is no indication of what factors they would notice. In the ICD dockets we screened there were no warning indications mentioned in the docket, and the investigators apparently did not ask for any. In the United Kingdom, the police are required to complete certain forms when they conclude that a person is at risk. They are required to tick a form that identifies risk from a possible seven categories including risk of harm to others. This form travels with the detainee, so that other police or prison officials have insight into that particular detainee's risk behaviour . It was also suggested that custody officers watch out for particular behavioural signals that indicate that detainees might have harmed themselves or could do so in the future. These include

  • direct expressions of intent (e.g. "I am going to slash my wrists")
  • expressions of failure, guilt or hopelessness
  • signs of past self harm behaviour, such as scars from previous attempts, and
  • acquisition of implements, such as secreting bits of cutlery.30

The ICD investigators felt that police often did not take such threats seriously, even where members of the family may have told the police of such a threat.

The SAPS Prevention of Torture Policy contains provisions relating to the search of persons in custody and the seizure of objects in their possession (Paragraph 9). In terms of this section, the community service centre commander must search, or authorise that a person in custody be searched, to ascertain what property he/she might have or have acquired which could be used for an unlawful or harmful purpose. That person can be searched – and intimate searches are provided for – and must be conducted in a way that has 'strict regard for decency'. A search should only be conducted to the extent that the community service centre commander considers necessary in the circumstances. Paragraph 9(5) says that articles of clothing may only be seized if there are grounds for believing that the person could use such articles to cause physical injury to himself or another person, damage to property, or assist in an escape. Dangerous objects found in the possession of the person in custody should be seized and held in safekeeping.

The SAPS Policy document does not stipulate exactly what could be deemed to be a dangerous object, or exactly what items of clothing should be removed. ICD investigators said that the police were often negligent in checking detainees for dangerous weapons before they were admitted to custody. An example was given of a man who attempted suicide by stabbing himself with his pocketknife – an item that ordinarily should have been removed. Another example is the case of 'E' in our docket analysis. The arresting police in this case had removed his service revolver, but had neglected to check whether he had any other firearms in his possession. He subsequently shot himself with his own firearm.

The police are supposed to conduct hourly visits to the cells. All of the police officers interviewed said that visits were conducted on an hourly basis for ordinary visits and half hourly for drunk or disturbed people. In most of the dockets where people had committed suicide the police records indicated that visits had taken place every hour on the hour. However, several ICD investigators spoken to doubted whether the visits were actually conducted and suspected that many police were merely filling in the visits on the hourly basis, particularly at night when they tend to avoid fulfilling that duty. In the case of one person who tried to commit suicide, the researchers were told that he waited until night time when the police did not visit and his cellmates were asleep. Indeed, it does seem surprising that each visit takes place exactly on the hour as is usually recorded in the Occurrence Book. The ICD investigator said that visits are usually only made when there is a change of shifts. It is only then that the police notice that something is wrong.

Independent oversight/visits at police stations

The Body of Principles for the Protection of All Persons under Any Form of Detention or Imprisonment says that

In order to supervise the strict observance of relevant laws and regulations, places of detention shall be visited regularly by qualified and experienced persons appointed by, and responsible to, a competent authority distinct from the authority in charge of the administration of the place of detention or imprisonment. (Principle 29)

According to the principles (Principle 29(1)), such a person should have the right to communicate freely and in full confidentiality with any detainee, subject to reasonable conditions to ensure security and good order.

There is no provision made in the SAPS Prevention of Torture Policy for independent visits to police cells. However, it does make provision for a person in custody to submit a complaint regarding torture directly to the Independent Complaints Directorate. If a complaint is made, it should be forwarded to the nearest Provincial Office of the ICD, and an Occurrence Book entry should be made to this effect (Paragraph 3(5)). However, given the ICD's present problems of capacity, it is unlikely that it would be able to respond to all complaints lodged from police cells.

South Africa has a history of abuse of people held in police custody, so that a system, which introduces a watchdog function over the police, should be seen as a necessity. Any system that relies solely on the lodging of complaints is less likely to pick up the systematic problems at police stations. Also, because independent visitors operate on a no-forewarning basis, their function is also a preventive one. A community-lay-visiting system is important for establishing trust between the community and the police.

The institution of community lay visiting schemes was piloted in South Africa 1993 and the Police Board drew up a policy in 1994. This scheme largely followed the English model of appointing independent community visitors. According to the policy, the purpose of the Community Visitor System is to 'enable members of the local community to independently observe, comment and report on the conditions under which people are held in police cells', and on the operation of various laws and regulations governing their welfare and ensuring accordance with fundamental human rights. It is uncertain to what extent these schemes are still operating, although it appears that they sometimes operate through the Community Police Forums

In terms of the South African Police Service Act, the SAPS must liaise with the community through CPF's with a view to:

  • establishing and maintaining a partnership between community and the service,
  • promoting communication,
  • promoting co-operation in fulfilling the needs of community policing,
  • improving transparency in the SAPS and accountability of the SAPS to the community, and
  • promoting joint problem identification and problem-solving (Section 18(1)).

According to interviews at police stations, at all but three of the police stations visited community visitor system was in operation. At one station, the system had been operating, but had recently ceased to function as a result of problems with the person from the CPF serving that function.

Police stations reported that they did receive visits from community visitors. These operated through the Community Police Forums. Visits varied from twice a month, to infrequent visits, and usually occurred after hours. One officer said that he would prefer more visits to be made, even daily.

Where these schemes did operate, the visitors have access to the cells, and are able to visit the people in custody. Usually they compiled a report setting out their findings, which is given to the Station Commissioner. They also met with the Station Commissioner occasionally. Most of the respondents viewed these schemes positively. At one station where the system had been operating for five years, it was said, 'Our Station Commissioner is very much in favour of these visits. He thinks it is a good control measure that they visit without warning.'

At another police station, most of the CPF members are reservists and therefore work with the cells on a regular basis. However, in this instance, there is possibly confusion of interests between those of safeguarding the public against crime, and protecting the rights of detained individuals.

Despite these reports, none of the ICD officials interviewed were aware of any lay visitor schemes.

It was reported that prior to 1994, police cells used to be visited by members of the International Red Cross and by Magistrates. The respondents said that these visits no longer occurred.

Investigations by the Independent Complaints Directorate

As soon as a death is reported to the ICD they try to go and investigate the scene. The police are instructed not to interfere with the scene, except in an attempt to resuscitate the person. It is preferable to have someone from the pathologist's office present at the scene, as well as someone to take photographs of the deceased – especially in instances where the deceased committed suicide.

However, it was said that not all police are co-operative with ICD investigations, and sometimes investigators suspect that crucial items of evidence are removed from the scene. They reported also that it was difficult to contact members of the police, or to obtain evidence or statements needed for the investigation. It was stated that there was better co-operation from senior rather than junior officials.

It was felt that not all police respect the ICD. However, one investigator said that the presence of the ICD was beginning to be felt amongst the police.

Section Three: Recommendations

A number of problems became clear through this study, some of which we believe could be redressed if the following recommendations were implemented.

1. Implementation of the SAPS Policy on Prevention of Torture and Treatment of Person in Custody

The drafting of this document was an important first step in heightening awareness of the rights of detained persons. It serves as a guide for police officers to use in the implementation of their daily work. Once the new Standing Orders are drafted, this should help to ensure implementation of the Policy. It is important that the new regulations or standing orders are finalized and made applicable as soon as possible.

It is of crucial importance that all police officers are made aware of the policy, and of its binding effect. Workshops should be held with police at station level advising them of the policy and of its implications.

2. Human rights training

It was evident in the study that few officers had received any training on human rights. This should become a priority for the police, as it does affect the work of every police officer on a daily basis. Knowledge of human rights is necessary in order for the police to fulfil their work within the boundaries of the law.

In Blaauw's study it was said, 'Crime-fighting police officers sometimes view police station lock-up not as a preliminary to punishment but as punishment in itself'.32 The ICD officials interviewed confirmed this view, stating also that police officials viewed detainees as criminals undeserving of human rights. It is thus important not only to create awareness of actual rights, but also to challenge the attitudes of police officers towards the human rights of offenders. This can be done through training and through changing the culture of the police.

3. Training on the management of people in custody

Most of the officers had received no training on the management of people in custody. Yet they are required to care for hundreds of detained suspects every month. This requires a certain amount of knowledge and skill, as it should not merely involve a lock-up function, particularly when it comes to dealing with the needs of particular categories of people – such as those who are mentally ill or suicidal. All police officers dealing with the detention and custody of people should receive training on the identification and management of people at risk. Training should also include searching prisoners for dangerous items, recording injuries, as well as reinforcing measures designed to ensure the security of the police lock up.

4. Allocation of specific officers to manage people in custody

The management of people in custody is a specific task with its own set of concerns, including that of ensuring the welfare of the detainee. However, a police official who is also tasked with the responsibility of crime fighting may find him/herself in somewhat of a conflictual role if also obliged to care for the needs of suspects. These kinds of conflicts can be reduced by making the role requirements more explicit, and by segregating conflictual roles.33 People should be specifically recruited and trained to perform the custodial function.

5. Management of people in custody who are not held in cells, and mechanical restraints

In three cases CSVR examined, people had been shot and killed by the police while attempting to escape. Although they were in custody of the police, they were not escaping from the cells. These cases were similar in that it appeared from our cursory examination that insufficient restraining measures had been taken to prevent escape. SAPS members need to be trained, or retrained on the importance of using effective, but humane restraints on suspects who are in transit or out on investigation. Appropriate use of restraining measures can be an important factor in the prevention of deaths in custody that occur when police try to stops a person from escaping. Failure to use restraints correctly should constitute a disciplinary offence, and appropriate action should be taken against members found to be infringing this provision.

The SAPS Policy on Prevention of Torture fails to deal with the use of restraints on suspects, or those in custody. In our view this is an oversight. Incorrect or inappropriate use of mechanical restraints could constitute cruel, inhuman or degrading treatment or punishment. An example of degrading treatment would be to bring a person before a court in leg-irons when it is unnecessary to do so. The Prevention of Torture policy should contain guidelines on when it is appropriate to use mechanical restraints on people, and what these restraint mechanisms should be. They should also contain guidelines on the maximum amount of time in which they can be used.

The use of restraints in cells is also an issue that should be dealt with in the policy. If mechanical restraints are used in the cells, the use thereof should be authorised by the Station Commissioner, and should only be used for the shortest time necessary.

A basic guideline should apply which stipulates that restraints may be applied by a member of the SAPS, as prescribed by regulation, only if it is necessary for the safety of a detainee or any other person, or for the prevention of damage to property, or if there is a reasonable suspicion that a detainee or prisoner may escape. Restraints should be generally confined to handcuffs.

6. Improving cell conditions

In the words of one police officer, detainees 'should be treated more like human beings'. Implicit in this suggestion is the improvement of the facilities in which detainees are accommodated so that their conditions are humane and in accordance with the requirements of the Bill of Rights. It would appear that the ongoing maintenance and upgrading of cells is a problem throughout the country. This should be viewed as one of the priority areas for expenditure.

Where the cells cannot be physically improved in the short term, station commissioners should ensure that they are made as habitable as possible, and that measures to reduce overcrowding are taken. Cleanliness and basic hygiene should be maintained at all times. Prisoners should be allowed a maximum of time outside of their cells in secure conditions.

Security measures should not be taken at the expense of humane conditions, so that security bars on the windows, for instance, should never interfere with the passage of fresh air.

7. Regular checks on detainees

Station Commissioners should ensure that regular checks are made to the cells, and that problematic detainees are monitored on a more frequent basis. Cell visits should also occur on a random basis, and not every hour, as prisoners are more likely to do an unlawful act or take their own lives once they are sure a visit is not likely to occur.

When conducting investigations, the ICD should ascertain whether regular cell visits have taken place. Prisoners should be interviewed to determine when they were visited. Disciplinary steps should be taken against police officials who have been found to be negligent in these duties.

8. Screening of 'people at risk'

The SAPS must develop procedures, or instruments to identify people who are at risk of harming themselves or others. They should also become aware of symptoms of illness or injury, and make careful note of them. Part of the screening process is placing greater emphasis on complaints or statements made by the detainee that would alert the police to this risk factor. Once these instruments have been developed, the police should be trained on their implementation.

These instruments should contain clear guidelines and procedures to be followed by those in charge of detainees should certain risk factors be identified. Selected police officers should receive training on counselling of suicidal detainees, or should be able to refer them to an appropriate agency for counselling. In addition, measures should be put in place to ensure that the people cannot harm themselves while in custody.

9. Thorough investigation

The ICD and the police need to conduct a thorough investigation into the circumstances of all deaths in custody. The version of the police should never be considered as sufficient, but supplementary statements or evidence should be obtained. The post mortem report should always be obtained.

Investigations by the ICD should not only be targeted towards establishing criminal liability, but should also be directed towards establishing how such deaths could be prevented in the future. In a study conducted for the ICD in 1998, Bruce argues that an approach that uses investigations for the purposes of 'professional review' may be preferable in that it is more conducive to cooperation by police officers with ICD investigators. He argues that it 'may be more effective in enabling investigations to address issues, and contribute to building a body of understanding, as to how custody/action deaths may be prevented'.34

In order to assist in the investigations, ICD investigators need to receive additional training and support.

10. Recommendations of the ICD to be followed up

In cases where recommendations have been made to the police, it is crucial that the ICD follows up to determine whether the recommended action has been taken. Even a finding of negligence on the part of the police should be seen in a serious light, and appropriate action taken against police officers.

The Station Commissioners should also be held to account for any death that occurs at their police station.

11. Prompt medical treatment

The police must ensure that prisoners receive prompt medical assistance when required. It was suggested by one ICD official that district surgeons be more available to assist prisoners with health problems and visit police cells on a regular basis.

12. Regular independent visits to police cells

Regular unannounced independent visits to police cells are an important oversight mechanism over the police. It is also an important means of establishing transparency and community faith that the police are not abusing their powers or infringing the human rights of those in their care. These regular visits can serve a preventative function through ensuring that policing policies are adhered to at station level. They can be an important source of information on the ongoing problems at police stations.

The ICD should establish communication and links with community visitor schemes where they exist, and should assist in the establishment of others. The ICD should itself also conduct regular visits to police cells to monitor and report on the conditions in custody.

13. Prevention of deaths resulting from injuries which occur before being taken into custody

A major problem is the large number of people who are injured, and who die as a result of injuries sustained before they are taken into custody. In most of the cases recorded by CSVR, injuries sustained prior to custody were mostly sustained at the hands of members of the community. In part, the causes of some of these deaths may be ascribed to community dissatisfaction with the criminal justice system and the perception that it is unable to bring perpetrators to book. A long-term preventative approach would be working towards making the criminal justice system more efficient in the apprehension of suspects, the investigation of cases, and the prosecution of offenders. In the interim, the SAPS have the responsibility of ensuring that persons who are injured through community vigilante action are removed to safety as soon as possible and receive medical treatment. This entails responding to reports of such assault as soon as they are logged, and arriving at the scene of the assault with whatever reinforcements are necessary to enable the police to rescue the individual.

But it is also important that the circumstances of the assault are investigated with the view to initiating a criminal investigation and prosecution. There is insufficient evidence in the ICD dockets to indicate whether or not this is happening. The ICD should also monitor whether the police are taking steps to prevent vigilante action in the communities.


Association for the Prevention of Torture (1999) Study: The impact of external visiting of police stations on prevention of torture and ill-treatment, Geneva.

Blaauw, E., Vermunt, R., and Kerkhof, A. (1996) Treatment in Police Custody: Organisational characteristics and performance of custodial divisions in Dutch police organisations. Criminal Justice and Behaviour, Volume 23, Number 4, page 553.

Bruce, D. (1997) Issues in the Investigation of Deaths in Police Custody or as a Result of Police Action, Centre for the Study of Violence and Reconciliation, unpublished paper.

Bruce, D. (1998). Towards a Stategy for Prevention: The occurrence of deaths in custody or as a result of police action in Gauteng, April – December 1997. Report for the Independent Complaints Directorate, Johannesburg.

Independent Complaints Directorate. (1997/8) Annual Report, CTP Book Printers. Pretoria.

Leigh, A., et al. (1998) Deaths in Police Custody: Learning the Lessons, Paper 26, Police Research Series, Police Research Group, London.

Oppler, S. (1999) Securing Cells: Reducing Collusion in Escapes from Custody, NEDCOR ISS Crime Index No. 1, 1999. Pretoria.

Oppler, S. (1999) Escapes from Police Custody: Is Government Winning the Game?, NEDCOR ISS Crime Index No. 1, Pretoria.

Policy Guideline (1994)s: Community Visitor System. The Police Board, South Africa.

Makhado, K., (20 August 1998). MEC Praises police over 'Fingers' arrest. Sowetan.

Nxusani, N., (20 August 1998) This is where it all ended … Sowetan.

Official Documents

South African Police Services Standing Orders, 1991 and 1992.

Regulations for the South African Police Services, No. R. 2086, 1996. South African Police Services Disciplinary Regulations.

South African Police Services (1999). Policy on the Prevention of Torture and The Treatment of Persons in Custody of the South African Police Service, Pretoria.

South African Police Service Act, No. 68 of 1995.


1 Note that if a person dies in custody but was injured as a result of a police action occurring outside of custody, there may be some overlap between a death 'which occurs in police custody', and a 'death which occurs as a result of police action'.

2 For further reading on the topic, see Bruce, D. Issues in the Investigation of Deaths in Police Custody or as a Result of Police Action, Centre for the Study of Violence and Reconciliation, unpublished paper, 1997.

3 Figures from the South African Police Service indicate 362 deaths occurred in police custody over the period January to December 1994, 195 for the period January to September 1995, and 210 as occurring in 1997. However each report used different categories, and only the 1995 figures provides a provincial breakdown. Source: Communication from the South African Police Service, Sub-Component Visible Policing to Centre for the Study of Violence and Reconciliation, 28 May 1999.

4 Independent Complaints Directorate. (1997/8) Annual Report, CTP Book Printers, at page 8.

5 The population figures are taken from the 1996 Census results. Sunday Times, October 25, 1998.

6 The statistics relating to the numbers of people arrested and held in police custody in each province was not available at the time of writing. However, statistics on the number of prisoners held in police cells in Gauteng police stations was available. A comparison between the number of deaths and the numbers in custody in Gauteng is reflected in Table 8.

7 The SAPS figures for 1995 (January to September) indicates that the greatest number of deaths occurred in Gauteng (48), followed by KwaZulu-Natal (46).

8 Interview with Steve Tiro: Director of Monitoring and Complaints Registry, ICD. 18 June 1999.

9 Three of these deaths occurred in South African prisons, and one in a Magistrate's court. In these cases the deceased was either a sentenced or un-sentenced prisoner. These cases have been included in this study as a reflection of what the ICD statistics include.

10 The case of Josiah 'Fingers' Rabotapi is reflected in ICD Complaint Control Number Register Gauteng Region as a death resulting from police action where the deceased was shot by 'other' firearm, and includes 'circumstances in which the victim does not die immediately after shooting, but lingers under medical care before expiring'. Media reports at the time of his death report that Rabotapi was in the custody of the police. He had been arrested and taken by the police to his apartment to point out cash and a firearm (Sowetan 20 August 1998). It was also reported that Rabotapi was shot by the police and died on the scene (Star, 21 August 1998).

11 Now referred to as the Director of Public Prosecutions.

12 Leigh, A., et al. Deaths in Police Custody: Learning the Lessons, Paper 26, Police Research Series, Police Research Group, London 1998.

13 A monkey chain is the kind of device used at hospitals to give patients support, or to raise injured limbs. It is attached to a bar above the bed.

14 In most police stations, the sentenced prisoners who, are required to perform labour for the police station, are held in separate cells referred to as the 'hard labour's cell'.

15 Leigh et al, at page 35.

16 In an article 'Another Dry White Season', Mail and Guardian, August 6 to 12 1999, the father of the deceased reports that he was not informed by the police why his son died in custody. He says that he saw the deceased's body in order to identify him, and said that he had been shot five times: including shots in the forehead, chin and leg. The father is quoted as saying that his toes were severely cut and bruised which he believes are signs of torture. He says that the ICD declined his request to provide him with a copy of the post-mortem report. The ICD representative is quoted as stating that the investigation is continuing. The police representative is quoted as saying that 'an official inquest has been opened'.

17 Bruce, D. and Komane, J. Police Attitudes to Informal Policing (Vigilante) Activities at Modal Interchanges, Draft report, Centre for the Study of Violence and Reconciliation, 1999, at 42.

18 Figures supplied by the SAPS Crime Analysis Information Centre, communication 20 May 1999.

19 Leigh, A., et al. Deaths in Police Custody: Learning the Lessons, Police Research Series, Paper 26, Police Research Group, London 1998, at iv.

20 SAPS Standing Order (G) 342 of 1991-05-02, 343, 345 (1992-07-08), 346- 348, 361, 363, 371 – 373, 381-382.

21 Blaauw, E. et al. 'Treatment in Police Custody: Organizational Characteristics and Performance of Custodial Divisions in Dutch Police Organisations', Criminal Justice and Behaviour, Vol. 23, No. 4, December 1996 at 554.

22 These were Krugersdorp, Kliptown, Kathlehong, Edenvale, Carltonville, Brixton, Brixton Murder and Robbery Unit, Springs, Sophiatown, and Protea Glen.

23 Section 35(1)(e) says 'Everyone who is detained … has the right to conditions of detention that are consistent with human dignity, including at least exercise and the provision at state expense, or adequate accommodation, nutrition, reading material and medical treatment'.

24 In terms of Section 29 of the Correctional Services Act, No. 8 of 1959, a person under the age of 14 years of age may be detained in a police cell or lock up after his arrest until he or she is brought to court for a period of 24 hours. A child who has attained 14 years but is under the age of 18 years may be held in a police cell for up to 48 hours. The presiding officer may authorise that children in the latter age group be detained for longer than 48 hours if he/she has reason to believe that his or her detention is necessary in the interests of the administration of justice and the safety and protection of the public and no secure place of safety within a reasonable distance from the court is available, or the person has committed one of the offences listed in the schedule to the Act. However, Ss (5)(a) states that such children may only be detained in a prison (but not a police cell or lock-up) if he/she is accused of having committed an offence or category of offence mentioned in Schedule 2 (the most serious offences) or in circumstances of such a nature which warrant such detention. Such a child shall be brought before court every 14 days to enable the court to reconsider the order of detention.

25 Oppler, S. 'Escapes from Police Custody: Is Government Winning the Game?', NEDCOR ISS Crime Index No. 1, 1999, page 20.

26 This is part of Project Five Star Initiative aimed at developing a 'Police Cells Facilities Improvement Guide' with procedures for replacing locks, and securing the cells and lock up areas. Electronic security systems are also part of this project, and are being piloted at Krugersdorp and Pietersburg police stations. Oppler, S. 'Securing Cells: Reducing Collusion in Escapes from Custody', NEDCOR ISS Crime Index No. 1, 1999, at 26.

27 Standing Order No. 361.22.1.3.

28 Paragraph 2(d)(iii) contains provisions which say that if a person is mentally handicapped or is suffering from a mental disorder, the community service centre commander should inform an appropriate adult of the grounds for his/her custody, and of the place where he/she is being detained, and ask that person to visit the detainee.

29 Leigh, A. Op Cit, at 14.

30 Op Cit, 15.

31 Policy Guidelines: Community Visitor System.

32 Blaauw, Op Cit, quoting Bittner, at 567.

33 Ibid.

34 Bruce. D., Towards a Strategy for Prevention: The Occurrence of Deaths in Police Custody or as a Result of Police Action in Gauteng, April – December 1997. Report for the Independent Complaints Directorate, Johannesburg, June 1998., at 25.

© Centre for the Study of Violence and Reconciliation

+ posts

CSVR is a multi-disciplinary institute that seeks to understand and prevent violence, heal its effects and build sustainable peace at the community, national and regional levels.

Related Content

TORTURE VICTIMS HAVE A RIGHT TO REHABILITATION: A Guide for Service Providers to Assist Victims of Torture in Securing their Right to Rehabilitation

Photo: Getty Images

The State Is Failing in Its Duty to Foster Inclusion and Safety of the LGBTIQ+ Community

Tunisia: The Colonial Legacy and Transitional Justice

The Smoke that Calls: Insurgent Citizenship and the Struggle for a Place in the New South Africa

Leave a Reply

Your email address will not be published. Required fields are marked *

Translate »