Data were collected on six large multigenerational pedigrees, four ascertained through a proband with major depression and two ascertained through a proband with a bipolar form of illness. Diagnoses were made using the SADS-L structured interview and Research Diagnostic Criteria (RDC). Complex segregation analyses were conducted on the bipolar and the major depression pedigree sets using a model allowing for both major locus and polygenic inheritance; in these analyses a variety of diagnostic schemes and assumptions concerning the lifetime population prevalence were examined. Linkage analyses on standard markers were conducted using parameters for transmission of susceptibility to illness derived from the segregation analyses. Results of the segregation analyses were quite sensitive to the diagnostic and prevalence assumptions. In the pedigrees ascertained through probands with a bipolar form of illness, we were unable to discriminate between major gene and polygenic inheritance. The data were compatible with Mendelian major gene transmission of susceptibility to illness when bipolar and schizoaffective manic diagnoses were considered as affected and the lifetime population prevalence was between 0.04 and 0.06. Outside this narrow prevalence range, or when additional diagnoses, such as major depression or hypomania, were included as expressions of liability to disease, major gene transmission of susceptibility to disease could be rejected. Similarly, in the pedigrees ascertained through probands with major depression, it was not generally possible to discriminate between major gene and polygenic transmission of susceptibility to illness. For a diagnostic scheme including only major depression as a manifestation of susceptibility to illness, there was a narrow range of lifetime population prevalence values (female prevalence ranging from 0.20 to 0.25, male prevalence set to 1/2 female prevalence) which yielded results compatible with major gene transmission. Linkage analyses for all markers yielded negative or inconclusive results. In one bipolar pedigree a lod score of 1.65 was found with a marker in chromosome 1 recommending further studies of this chromosome.