Investigators continue to debate whether the Dexamethasone Suppression Test (DST) reflects clinical severity or degree of melancholia ("endogeneity"). To evaluate this question, we studied 73 drug-free inpatients diagnosed with Schedule for Affective Disorders and Schizophrenia/Research Diagnostic Criteria (SADS/RDC) as having major depressive disorder (MDD). We compared absolute and dichotomous DST values (DST suppression versus nonsuppression) with absolute and dichotomous measures of endogeneity (as measured by operationally defined RDC items) and with Hamilton Rating Scale for Depression (HRSD) scores that were collected immediately prior to treatment. We found that degree of endogeneity correlated moderately (r = 0.27) but significantly (p = 0.02) with absolute DST values; DST nonsuppression increased proportionately with changes in categorical endogenous subtype (0% of the nonendogenous patients were nonsuppressives, 52% of probable endogenous, and 61% of subjects definitely endogenous); mean values for maximum DST concentrations increased steadily with categorical endogeneity (nonendogenous, 1.44 microgram/dl; probable endogenous, 7.65 micrograms/dl; definite, 10.93 micrograms/dl; p = 0.01); HRSD scores correlated more strongly (r = 0.45, p = 0.000) with maximum DST levels than did the degree of endogeneity. Age and weight changes did not account for the relationship of endogeneity to DST values. These data suggest that maximum postdexamethasone plasma cortisol levels reflect overall severity of depression and endogeneity and that endogeneity per se is highly confounded with severity.