Subjects who meet the criteria for an affective syndrome possibly are aetiologically heterogeneous. An approach to this possibility involves examining affectively ill subjects obtained by different methods of ascertainment. This study compares depressed and manic subjects who are related to affectively ill probands with affectively ill subjects who were obtained from a study of a control population, and, therefore, were less likely to be familial. The subjects were identified in a large collaborative study of depression where both family members as well as controls were personally interviewed and followed up for 6 years after admission to the study. Data were obtained on subtypes of affective disorder using the Research Diagnostic Criteria and information was gathered about psychiatric hospitalizations, suicide attempts, alcoholism and psychological functioning prior to admission. Similar assessments were made for the comparison groups for the 6 year period between intake and follow-up. Relatives of bipolar I/schizoaffective manic probands were more likely to show mania than affectively ill controls or relatives of unipolar/schizoaffective depressed probands. Affectively ill controls were less likely to be hospitalized and less likely to suffer from an incapacitating depression. They were also likely to have functioned in a more healthy fashion than the affectively ill relatives of the bipolars and unipolars, in the 5 years before admission to the study. In the 6 year follow-up, both the subjects themselves and raters assessed the depressed controls as functioning better than the affectively ill relatives of the probands. Further, assessment of global adjustment during the 6 year period was worse for the relatives of affectively ill probands than for the depressed controls. Length of major depression was longer in relatives of bipolar and unipolar probands than in controls. Though all of the subjects in this study met research criteria for an affective illness, there were marked differences in the qualitative aspects of these illnesses with the relatives of affectively ill probands, who functioned less well and had longer and more severe episodes and more hospitalizations.