Survival analysis of familial covariates of risk for affective illness demonstrated a significant effect on the proband's diagnosis when affection status included bipolar illness, but not major depressive disorder. A cohort effect was indicated only when broad spectra (including bipolar II and major depressive disorder) of illness were defined. Multifactorial analyses of familial correlations for affective illness evidenced neither sex-specific correlations nor prevalences. We suggest the examination of risk variables prior to undertaking familial analysis.