A sample of 184 hospitalized primary unipolar depressives was used to examine the relationship between the dexamethasone suppression test (DST) and Winokur's familial typology as a function of the broadness of the criteria used to diagnose the patients' first-degree relatives. When a high threshold was used (ie, alcoholic relatives were treated, antisocial relatives had legal problems, and depressed relatives were hospitalized), the DST-familial subtype relationship was much stronger than when a low threshold was used (ie, relatives were diagnosed according to the Family History Research Diagnostic Criteria). The differences in diagnosing family members may be responsible for the inconsistent results reported by investigators examining the relationship between Winokur's typology and the DST.