Seventy patients fulfilling DSM-III criteria for major depression were given the 1.0 mg overnight dexamethasone suppression test, with salivary cortisol concentrations being measured as the dependent variable. Using both the DSM-III and the Research Diagnostic Criteria, we categorized the patients into four groups based on increasing frequency of endogenous symptomatology. Among these four groups there were no significant differences in salivary cortisol concentrations either before dexamethasone or eight, 16, and 24 h after dexamethasone. Similarly, there were no significant differences among the groups in either the ratios of post- to pre-dexamethasone salivary cortisol or the frequencies of positive tests based on several criterion levels of cortisol for the three post-dexamethasone samples. Multiple regression analyses indicated that the Hamilton depression rating scale item "somatic anxiety" was significantly negatively related to post-dexamethasone cortisol concentrations. We conclude that, for our sample of major depressives, the salivary cortisol dexamethasone suppression test showed no utility as a laboratory correlate of depressive episodes with endogenous features.