Four hundred and fifty directly interviewed relatives of probands with non-psychotic unipolar major depression and 320 directly interviewed relatives of controls were compared by the prevalences of personality disorders (P.D.) as defined by DSM-III-R, in relation to presence or absence of the relatives' affective disorder. Overall, there was only a trend for an increased risk for P.D. in relatives of depressed patients. However, P.D. and unipolar major depression co-occurred more frequently in relatives than expected by chance. It is suggested that this association is mainly due to non-familial factors. Compared with other P.D., the relationship of borderline P.D. to major depression was not substantially stronger.