Résumé : The validity of the dexamethasone suppression test was evaluated for the differential diagnosis of primary and secondary depressions. Forty hospitalized psychiatric patients (14 primary depressed, 15 secondary depressed and 11 non-depressed controls) were studied. The Research Diagnostic Criteria of Spitzer et al. (1978) were used to classify these patients. Eight out of the 14 primary depressed patients had an abnormally high plasma cortisol at 4 p.m. after dexamethasone. Only 2 out of the 15 secondary depressed patients and none of the 11 controls had an abnormal response to dexamethasone. Based on these results, the dexamethasone suppression test has a sensitivity of 57%, a specificity of 87% and a predictive value of 80%. The determination of urinary free cortisol excretion does not improve the performance of the test. © 1981.