OBJECTIVE The purpose of this study was to determine 1) the reliability of diagnoses of schizophrenia at coronal autopsy and 2) the degree to which the use of different diagnostic instruments for schizophrenia would affect postmortem brain research. METHODS Eighty-three subjects, recorded at coronal autopsy as having had schizophrenia, were referred for neurochemistry studies. The diagnoses reported to the state coroner's office were reevaluated by a review of psychiatric case histories by clinicians using semistructured assessment and diagnostic criteria. RESULTS The application of DSM-III-R, Research Diagnostic Criteria (RDC), ICD-10, Schneiderian, and Feighner criteria to the diagnosis of the 83 subjects revealed that 63.9%, 48.2%, 48,2%, 43.4%, and 42.2%, respectively, met the criteria for schizophrenia. Highest concordance was between the RDC and ICD-10 systems, while lowest concordance was between the RDC and Schneider systems. CONCLUSIONS These data suggest that unless carefully reviewed, diagnosis may be a major confounding factor in postmortem studies of brain tissue from subjects with schizophrenia.