Objective:To determine the characteristics of victims killed by firearms, whether they died by suicide, homicide or accident, and the types of firearms involved.Design:A retrospective and descriptive study.Setting and subjects:Coronial autopsy data were obtained from the Brisbane Laboratory of Pathology and Microbiology. All 587 firearm deaths during the period 1980–1989 were included.Main outcome measures:The proportions of deaths by suicide, homicide, accident or undetermined cause. In cases of homicide, we sought to define the relationship of the perpetrator to the victim.Results:The types of death were suicide 76%, homicide 18%, accidents 3% and undetermined 3%. Most homicides were of a domestic nature as opposed to crime. Small calibre rifles were the most common weapon; handguns were rarely used.Conclusions:Recent debate regarding firearm legislation has failed to recognise the predominance of interpersonal and psychological issues in the causation of firearm deaths. Crime is a motive for only a minority. Medical practitioners may reduce firearm deaths both by organisational representation and individual efforts in clinical practice. In particular, attention should be given to the identification of individuals or families at risk, the detection and treatment of psychological disorder and advice regarding firearm disposal.