This study examined the validity of the family history method for diagnosing schizophrenia, schizophrenia-related psychoses, and schizophrenia-spectrum personality disorders in first-degree relatives of schizophrenia probands. This is the first large-scale study that examined the validity of the family history method for diagnosing DSM-III-R personality disorders. The best estimate DSM-III-R diagnoses of 264 first-degree relatives of 117 adult-onset schizophrenia probands based on direct structured diagnostic interviews, family history interview, and medical records were compared to Family History Research Diagnostic Criteria (FH-RDC) diagnoses based on the NIMH Relative Psychiatric History Interview and to family history Structured Clinical Interview for DSM-III-R: Personality Disorders (SCID-II) diagnoses based on the SCID-II adapted to a third person format. Diagnoses of relatives were made blind to proband diagnostic status. The median sensitivity for schizophrenia and the related psychoses was 29% (range 0-50%), the median specificity 99% (range 98-100%), and the median positive predictive value (PPV) 67% (range 20-80%). The median sensitivity for the personality diagnoses was 25% (range 14-71%), the median specificity 100% (range 99-100%), and the median PPV 100% (range 67-100%). The family history method has low sensitivity but has excellent specificity and PPV for schizophrenia, schizophrenia-related psychoses, and schizophrenia-spectrum personality disorders. The kappa coefficient for the family history method was moderately good for the psychoses (0.598) and for paranoid and schizotypal personality disorder (0.576). Using the family history method, the validity of making schizophrenia-related personality disorder diagnoses was comparable to that of making psychotic disorder diagnoses.