This report examines the reliability, concordance, and long-term stability of the subtypes of schizophrenia defined by four major diagnostic systems (DSM-III, DSM-III-R, ICD-10, and Tsuang-Winokur criteria) and rated both for the first hospitalization and for a best estimate diagnosis reflecting lifetime evolution of symptomatology. Schizophrenics studied belonged to two samples of multiply affected families, namely a sample selected in France and a sample of non-metropolitan French identified in the island of La Réunion. ICD-10 and DSM-III-R show opposite stringency regarding subtyping of schizophrenia, with DSM-III-R having a narrow and ICD-10 a broader definition of specific subtypes. Long-term stability of each subtype was fairly good, stability being the highest for hebephrenics and only intermediate for paranoid and undifferentiated subtypes. Comparison of two different cultural and geographical regions reveals an overall similarity of subtype frequencies in familial schizophrenia. The implications of the results for the choice of diagnostic procedures in family studies of schizophrenia are discussed.