We studied a sample of 26 unipolar and 134 bipolar probands for a selection of familial, demographic, clinical and laboratory variables. We found a high proportion of unipolar illness in the relatives of bipolar probands and bipolar illness in the relatives of unipolar probands, inconsistent with the present dichotomous classification of affective disorders. The two groups were also similar on a number of clinical and demographic variables, and displayed equal amounts of cortical dysfunction as measured by electroencephalographic and neuropsychological techniques. The groups differed significantly on only 2 variables, with unipolars having a greater proportion of females and a later onset age than bipolars. We interpret these findings to be suggestive of homogeneity rather than heterogeneity for the more severe forms of affective disorders and offer several alternatives to the present unipolar/bipolar dichotomy which require further testing.