In a family study involving 139 probands with DSM-III-R catatonic schizophrenia and 543 first-degree relatives, we investigated age-specific morbidity risk according to Leonhard's clinical distinction between systematic and periodic catatonia. This dichotomy is based on different types of symptomatology, course, and outcome. In systematic catatonia the age-corrected morbidity risk was 4.6%. In periodic catatonia, however, there was an age-corrected morbidity risk with homogenous psychoses of 26.9%, and more parents than siblings were affected. This points strongly to a major gene effect in periodic catatonia. Furthermore, a pairwise comparison of patients and their parents revealed patterns of anticipation, i.e., the probands' age at the onset of disease was significantly earlier than that of their parents (P < 0.001). Similarly, anticipation was apparent in pedigrees with three successive generations affected. This inheritance pattern with homogenous psychoses and anticipation indicates that genes with trinucleotid repeat expansion or other repetitive elements affecting gene expression may be involved in the etiology of periodic catatonia. Thus, periodic catatonia as a specific clinical subtype of schizophrenia is a promising candidate for molecular genetic evaluation.