OBJECTIVE: The authors analyzed data from a family sample ascertained for a genetic linkage study of bipolar disorder to address the following questions: Do the major clinical subtypes of familial affective disorder have distinct distributions of age at onset? What factors other than clinical subtype affect these distributions? After controlling for these factors, do the differences in age at onset persist among the subtypes? METHODS: Eighty-two families were ascertained through a treated proband with bipolar disorder who had a family history of two or more affected siblings or one affected sibling and one affected parent. After participating in an interview conducted by a psychiatrist using the Schedule for Affective Disorders and Schizophrenia--Lifetime Version, 274 probands and their first-degree relatives were diagnosed as having bipolar I, bipolar II, or recurrent unipolar disorder according to Research Diagnostic Criteria. Age at first major affective episode and other clinical data were collected. RESULTS: Onset age distributions were similar for bipolar I and bipolar II disorder but significantly different for recurrent unipolar disorder. This finding persisted after adjustment for a significantly earlier onset among females. Subjects with affective disorder in both parental lines (bilineal) also experienced a significantly earlier onset. Substance abuse, physical illness, and sex of the affected parent had no significant impact on onset age. CONCLUSIONS: Although differences in age at onset may reflect several factors, these results provide indirect support for the view that bipolar I and bipolar II disorders are genetically related phenotypes and suggest that bilineal families may be more complex than previously assumed.