We examined familial rates of affective disorder and related illness in a cohort of 955 probands studied at five centers in the National Institute of Mental Health Collaborative Study of the Psychobiology of Depression: Boston, Chicago, Iowa City, New York, and St. Louis. Six hundred sixteen of these probands were entered into a family study, and 3423 of their first-degree relatives were evaluated. The probands were divided into five diagnostic groups: schizoaffective-bipolar (n = 37), schizoaffective-depressed (n = 18), bipolar I (n = 151), bipolar II (n = 76), and unipolar (n = 330). The relatives of bipolar I probands had a higher rate of bipolar I illness than the relatives of unipolar probands, but the relatives of unipolar probands did not have a higher rate of unipolar illness than the relatives of bipolar I probands. The relatives of probands with schizoaffective disorder, depressed subtype, had a higher rate of schizophrenia than the relatives of schizoaffective-bipolar probands, suggesting that bipolar schizoaffective disorder may be closer to pure affective disorder while schizoaffective depression may be closer to schizophrenia. An increase in bipolar II illness was also observed in the relatives of bipolar II probands. Overall, these data support the widely accepted distinction between bipolar and unipolar affective disorders.