The psychological autopsy method offers the most direct technique currently available for examining the relationship between particular antecedents and suicide. This systematic review aimed to examine the results of studies of suicide that used a psychological autopsy method. A computer aided search of MEDLINE, BIDS ISI and PSYCHLIT, supplemented by reports known to the reviewers and reports identified from the reference lists of other retrieved reports. Two investigators systematically and independently examined all reports. Median proportions were determined and population attributable fractions were calculated, where possible, in cases of suicide and controls. One hundred and fifty-four reports were identified, of which 76 met the criteria for inclusion; 54 were case series and 22 were case-control studies. The median proportion of cases with mental disorder was 91% (95 % CI 81-98%) in the case series. In the case-control studies the figure was 90% (88-95%) in the cases and 27% (14-48%) in the controls. Co-morbid mental disorder and substance abuse also preceded suicide in more cases (38%, 19-57%) than controls (6%, 0-13%). The population attributable fraction for mental disorder ranged from 47-74% in the seven studies in which it could be calculated. The effects of particular disorders and sociological variables have been insufficiently studied to draw clear conclusions. The results indicated that mental disorder was the most strongly associated variable of those that have been studied. Further studies should focus on specific disorders and psychosocial factors. Suicide prevention strategies may be most effective if focused on the treatment of mental disorders.