About the National Deaths in Custody Program

In 1987 the Australian Government established the Royal Commission into Aboriginal Deaths in Custody (RCIADIC) in response to concerns about the number of Aboriginal and Torres Strait Islander people dying in custody. The RCIADIC found that Indigenous people were not more likely to die in custody than non-Indigenous people but were significantly over-represented in custody. Recent figures show that the over-representation of Indigenous people in custody remains a problem in Australia (ABS 2017).

The RCIADIC also noted the lack of reliable statistics available on deaths in custody in Australia. To fill this information gap, the RCIADIC recommended the establishment of a program to monitor Indigenous and non-Indigenous deaths in prison, police custody and youth detention (recommendation 41, RCIADIC 1991). In response, the National Deaths in Custody Program (NDICP) was established at the Australian Institute of Criminology in 1992. The final report of the RCIADIC outlined the types of deaths that would require notification to the NDICP (recommendation 41, RCIADIC 1991). They are:

  • a death, wherever occurring, of a person who is in prison custody, police custody or youth detention;
  • a death, wherever occurring, of a person whose death is caused or contributed to by traumatic injuries sustained, or by lack of proper care, while in such custody or detention;
  • a death, wherever occurring, of a person who dies, or is fatally injured, in the process of police or prison officers attempting to detain that person; or
  • a death, wherever occurring, of a person attempting to escape from prison, police custody or youth detention.

As the NDICP was established solely to monitor deaths that occur in prison, police custody and youth detention, it does not consider deaths in immigration detention centres.

Compiling the National Deaths in Custody database

The information held in the NDICP database is based on two main data sources:

  • NDICP data collection forms completed by all state and territory police services and corrective departments in Australia; and
  • coronial reports, such as transcripts of proceedings and findings, as well as toxicology and post-mortem reports.

NDICP data collection forms allow information to be recorded on approximately 80 variables relating to the circumstances and characteristics of each death. Coronial data are accessed through the National Coronial Information System and are primarily used to confirm the cause and manner of death.


The definitions used to determine whether an individual’s death should be included as a death in custody are based on recommendations of the RCIADIC (see Box 1). The definition of a death in police custody was expanded following a resolution of the Australasian Police Ministers Council to include other close contact deaths that occurred during police operations (Category 1b) and deaths that occurred during custody-related police operations that did not involve close contact with the deceased (Category 2). Prior to this change, only deaths in institutional settings were recorded (Category 1a).

Deaths occurring in the following circumstances are excluded from the NDICP:

  • deaths during rescue-related police operations such as search and rescues and evacuations;
  • deaths during police operations to prevent a suicide attempt, where the purpose was not to detain the deceased due to a breach of the law; and
  • deaths of innocent bystanders in police operations, including pedestrians or vehicle passengers.

Box 1: Defining deaths in custody

Deaths in prison custody

Deaths in prison custody include deaths that occur in prison or youth detention facilities. This also includes the deaths that occur during transfer to or from prison or youth detention centres, or in medical facilities following transfer from adult and youth detention centres (RCIADIC 1991: 189–190)..

Deaths in police custody

Deaths in police custody are divided into two categories:

Category 1

1a: Deaths in institutional settings (eg police stations or lockups, police vehicles or hospitals, during transfer to or from such institutions, or following transfer from an institution).

1b: Other deaths in police operations where officers were in close contact with the deceased. This would include most raids and shootings by police. However, it would not include most sieges where a perimeter was established around a premise but officers did not have such close contact with the person that they were able to significantly influence or control the person’s behaviour.

Category 2

Other deaths during custody-related police operations. This would cover most situations where officers did not have such close contact with the person to be able to significantly influence or control the person’s behaviour. It would include most sieges and cases where officers were attempting to detain a person—for example, during a pursuit.

Borderline cases

The NDICP uses the definition of a death in custody recommended by the RCIADIC as a guide to which cases should or should not be included in the NDICP database. While most cases are straightforward and fall within the definition, every year there are some cases where it is unclear whether the death should be classified as a death in custody.

For the purposes of the NDICP, a person is considered to be in custody when they are not free to leave the detention or arrest of police or corrections officials. As outlined in the definitions provided in Box 1, this includes deaths that occur in a hospital if the injuries or illness suffered while in custody caused or contributed to that death. In cases where police were clearly in the process of detaining or attempting to detain a person immediately prior to death, such as in shootings, sieges, raids and pursuits, the person is considered to have been in custody at the time of death. In all of these cases, the question of inclusion centres on whether the deceased was in custody at the time of death.

Any borderline cases are excluded from analysis pending their coronial outcome. This may result in a delay of several years to collect data regarding those particular borderline cases, as they may not be heard in their jurisdiction’s coroner’s court for months or years. Despite this, relying on coronial decisions ensures the integrity and reliability of the NDICP over the longer term, as coronial findings are legally binding determinations based on all available evidence. It is important to note that this means the total numbers of deaths may be revised in the future as the inclusion or exclusion of borderline cases is resolved.

Period of data collection

Data for deaths in prison custody are presented from 1979–80, while data for deaths in police custody are presented from 1989–90.

Trend data may fluctuate markedly due to the small numbers of deaths in custody. Therefore, caution should be exercised with interpretation.

Cause versus manner of death

The NDICP collects information on both the cause and manner of each death. Cause of death information relates to the direct cause of death as reported by the coroner or by police and prison authorities. Manner of death is a related variable, but it refers to the accountability or responsibility for the death as reported by the coroner or by police and prison authorities. Therefore, in some cases cause and manner will correspond; for example, where a person dies as a result of natural causes, their death will be recorded as natural causes for both cause and manner of death. In other cases, cause and manner of death will differ; for example, where a person dies after hanging themselves, the cause of death will be recorded as hanging and the manner of death will be recorded as either self-inflicted or accidental hanging.

Calculating death rates

Where rates are presented (unless otherwise stated), they have been calculated using the annual Prisoners in Australia results of the National Prisoner Census (ABS 2017), which counts all prisoners who were in legal custody in each jurisdiction as at midnight, 30 June. Further, where trends in rates of death are presented, the rates are calculated back to 1982, as prison census data are not available prior to 1982.

Rates of police custody deaths are not presented because there is no reliable data source for the numbers of people who are placed into police custody each year and the numbers of people who come into contact with police in custody-related operations.

AIC research

Trends and Issues in Crime and Criminal Justice

Statistical Reports

Research Reports

Statistical Bulletins