In family study concentrating on 139 probands with chronic DSM-III-R schizophrenia, catatonic type, 83 probands (41 women, 42 men) met the criteria for periodic catatonia and 56 probands (14 women, 42 men) for systematic catatonia according to the Leonhard classification. The reliability and stability of this subclassification were tested by 2 experienced psychiatrists working independently of each other. Both diagnosticians were kept blind as to the probands' family history. The 139 probands had a total of 543 first-degree relatives. Only those hospitalized for schizophrenia were allocated to the group of afflicted family members. Diagnostic reliability was kappa statistic 0.93 and diagnostic stability during catamnesis reached 97% and kappa of 0.93. Life-table analyses revealed that the age-corrected risks were significantly different in periodic and systematic catatonia. In systematic catatonia mothers had a risk of 6.8%, fathers 2%, and randomly selected sibs 3%. IN periodic catatonia an excess of homologous psychoses was apparent: There was a risk of 33.7% for mothers, 15.4% for fathers, and 24.4% for sibs. The quota of afflicted parents (33 of 161) was higher than that of sibs (26 of 162). In periodic catatonia, 59% of the families were multiple afflicted with pronounced unilineal vertical transmission. In 10% of the families 3 successive generations suffered from the disease and were treated in hospital. The results of the study led to the following hypotheses: Periodic and systematic catatonia are valid subgroups of DSM-III-R schizophrenia. In systematic catatonia heritability is very low. Periodic catatonia is a familial disorder. Homogeneity of familial psychoses and unilineal vertical transmission with anticipation are consistent with a major gene effect. Periodic catatonia seems to be a promising candidate for molecular genetic evaluation.