鈥� Analysis of family history and antidepressant drug response variables of 100 "neurotic" depressives followed up prospectively over three to four years disclosed that primary depressions (unipolar and bipolar) could be distinguished from nonprimary cases by (1) the early occurrence of "pharmacological-hypomania;" (2) family history of bipolar illness; (3) family history for affective disorder in two or three consecutive generations, especially when "loaded." Although each of these variables alone occurred in only one fifth to one third of the primary group, they individually displayed better than 95% specificity for it. Thus, the confidence with which the diagnosis of primary affective illness could be made in the presence of any of these variables ranged from 88% to 100%. These findings argue for considering such nonsymptomatological variables for their potential in strengthening the phenomenologic diagnostic criteria for depressive illness.