Objective: Most investigators presume that schizophrenia and affective disorder are separate diseases. Others have proposed alternatives to this Kraepelinian view. These alternatives were addressed by preliminary analyses of data from a family study of psychopathology. Method: The authors identified 1,895 first-degree relatives of 166 patients with DSM-III schizophrenia, 71 patients with affective disorder, and 85 medical comparison probands; 949 relatives were blindly diagnosed. Results: The risks for schizophrenia and affective disorder (unipolar melancholia and bipolar disorder combined) were significantly higher in the relatives of the schizophrenic probands and the relatives of the probands with affective disorder than in the relatives of the comparison probands. The morbid risk for nonmelancholic depressions was not significantly higher. Among the relatives of the schizophrenic probands, the risk for affective disorder was highest among the relatives of the patients with 'core' schizophrenia, who were younger at illness onset, had chronic illness, had severe emotional blunting, and showed few affective features. Conclusions: Despite limitations, these preliminary analyses, consistent with other studies, suggest some familial relationship between schizophrenia and severe forms of affective disorder.