These analyses focused on the relationship of affective disorders to HLA. The sample of 86 previously unpublished families from Ontario and 10 unpublished pedigrees from Newfoundland, as well as the original Weitkamp et al. [1981] sample of 20 families, the samples provided by NIMH [Goldin et al., 1982] (18 families), and Egeland et al. [1987] (2 large pedigrees), were examined using sib pair methods as well as standard linkage analysis of the full pedigrees. A variety of genetic models were examined. There was no evidence for linkage from any of the samples based on either analytic approach or for any genetic model. Groups of sibling pairs concordant for affective illness, concordant for being well, or discordant for affective illness did not differ in the proportion of genes identical by descent. No group differed significantly from 50%. From the analysis of full pedigrees, linkage to HLA could be excluded to a distance of approximately 20 to 25 centimorgans. There was no evidence for linkage heterogeneity. These results argue against linkage of affective disorder to HLA.