A logistic model was used to analyze the pattern of affected relatives of probands with primary affective disorders (PAD). The sample consisted of 242 patients, diagnosed as either unipolar (UP, 107) or bipolar (BP, 135) and 430 control nonpsychiatric inpatients and all first degree relatives of both groups. Age correction was applied to both groups. The analysis showed a significant baseline increase in frequency of PAD among relatives of PAD probands with siblings more likely to be affected than parents. The difference in frequency of PAD according to sex of relative almost reached significance. Specific diagnosis (UP or BP) of proband did not significantly affect the probability of relatives becoming ill. Genetic models incorporating sex-specific thresholds were able to explain the data satisfactorily as resulting from either Single-Major-Locus inheritance or Multifactorial-Polygenic inheritance.