Psychological autopsies are an important research tool in establishing risk factors associated with suicide. We report the results of a validity study comparing psychological autopsy-generated DSM-III-R diagnoses in suicides and non-suicides with chart diagnoses generated by clinicians who had treated the subjects prior to death. The Structured Clinical Interview for DSM-III-R Disorders (SCID-P) and the Structured Clinical Interview for DSM-III-R Personality Disorders (SCID-II) were used to make independent post-mortem diagnoses. Comparison of research diagnoses with clinician ante-mortem diagnoses generated kappa coefficients of 0.85 for Axis I diagnoses and 0.65 for Axis II conditions. These kappa coefficients compare favourably with direct patient interview reliability studies. This provides evidence for the validity of the psychological autopsy as a method of determining psychiatric diagnosis.