Data from a family study of bipolar affective illness in New York (1968-1972) are reexamined from the vantage point of X-linked dominant inheritance and genetic heterogeneity. For comparison, published data from a similar study in Bethesda (1974-1979) and from a literature summary are also re-analyzed. The New York data support the presence of a change in rates of affective illness over time (secular trend), with a substantial increase in risk at early ages for individuals born after 1920. Hence, risk ratios by sex of proband (male vs female) are calculated for each type of relative, thereby controlling simultaneously for generation and sex of relative. The risk ratios for all three studies are in the direction predicted by X-linkage. In addition, logistic regression is performed, incorporating the effects of generation, sex of relative, sex of proband, and X-linkage. In all three samples, the largest effect is due to X-linkage, which reaches statistical significance in the New York and literature summary samples. The results also suggest genetic heterogeneity, in that only a subgroup of bipolars carry the X-linked gene. Crude estimates predict that perhaps one-third of bipolars carry the X-linked gene. Analysis of the New York material also indicates that an X-linked subgroup may be associated with early onset (before 30 years old). These results are consistent with those from linkage analysis with markers on the long arm of the X chromosome.