STUDY OBJECTIVE--The aim was to determine whether clustering of suicide occurred in prisoners and detainees in police cells and to describe the characteristics of any clusters identified. DESIGN--Mortality from suicide was examined for evidence of clusters. SETTING--Prisons (1971-1988) and police cells (1980-September 1991) in New Zealand. SUBJECTS--82 male suicides. MAIN RESULTS--There were 38 male suicides in prisons and 44 in police cells over the time periods. Due to the increasing trend in prison suicides, the 1971-1982 and 1983-1988 time periods were examined for evidence of clusters separately. For suicides in police cells the time periods studied were 1980-1982, 1983-1988, and 1989-1991. The event of suicide was treated as a Poisson process with all suicides in a time period used to calculate the mean. Four suicide clusters occurred between 1985 and 1988 in prisons. These did not explain the increased suicide rate over this period. Three clusters of suicide in police cells occurred between 1980 and 1991, which had different characteristics from suicide in prisons in that these three clusters were of longer duration. CONCLUSIONS--The different characteristics of prison and police cell clusters of suicide indicated separate transmission of the contagious component of suicide between these groups. Distinct strategies to reduce clusters of male suicide in custody are required for prisoners and those held in police cells.