Whereas a growing body of evidence suggests that cycloid psychoses have to be separated from schizophrenic psychoses, their relations to bipolar affective disorder are less clear. To further clarify this issue a controlled family study was undertaken. All living and traceable adult first-degree relatives of 45 cycloid psychotic, 32 manic-depressive and 27 control probands were personally examined by an experienced psychiatrist blind to the diagnosis of the index proband. Data about not traceable relatives were collected by the "Family-History"-Method. A catamnestic diagnosis was established for each of the 431 relatives blind to family data. Age-corrected morbidity risks were calculated using the life-table method. Relatives of cycloid psychotic patients showed a significantly lower morbidity risk for endogenous psychoses in general and manic-depressive illness compared to relatives of patients with manic-depressive illness. The familial morbidity risk for cycloid psychoses was low and did not differ significantly in both proband groups. Relatives of cycloid psychotic patients however did not differ significantly from relatives of controls regarding familial morbidity. Our time-consuming methodical procedure implicated a relatively small number of participants due to restricted personnel resources. The restriction to hospitalised probands could possibly cause a limited representativity of the study sample. Our results suggest that cycloid psychoses are aetiologically different from manic-depressive illness and could not be integrated into a spectrum of bipolar affective disorders. The findings provide further evidence for a nosological independence of cycloid psychoses.