Professional raters who were blind to proband diagnosis used the schedule for affective disorders and schizophrenia (SADS-L) and the Research Diagnostic Criteria (RDC) to evaluate 1,210 first-degree relatives of 327 probands with primary major depression, participating in the family sub-study of the NIMH Collaborative Study of the Affective Disorders--Clinical Branch. Bipolar II probands were significantly more likely to have bipolar II relatives than were non-bipolar or bipolar I probands. Bipolar II probands were slightly more likely than non-bipolar probands and slightly less likely than bipolar I probands to have relatives with bipolar I illness. Similar patterns have emerged in two other recently reported family studies of bipolar II illness. Taken together, these data suggest heterogeneity among patients with bipolar II depression. Some appear to be genotypes for bipolar I illness, while a small proportion may be genotypes for non-bipolar illness. A third group, of undetermined size, may breed true.