This prospective study was conducted in order (1) to examine which postdexamethasone cortisol value -i.e., 8 a.m., 4 p.m. or peak cortisol – is most suitable as a laboratory test to help confirm the diagnosis of melancholia and (2) to investigate the influence of the dexamethasone levels in the results of the dexamethasone suppression test (DST). To this end we administered the DST to 48 controls and 115 depressed inpatients categorized according to DSM-III. The 8 a.m. and 4 p.m. dexamethasone levels were determined in 100 subjects. We found that an 8 a.m. postdexamethasone cortisol value ≧3.5 µg/dl was of the most significant diagnostic value in order to separate melancholia from normal controls and/or minor depressives. The 8 a.m. and 4 p.m. dexamethasone values did not differ between healthy controls, minor and severely depressed patients. Although cortisol nonsuppressors exhibited significantly lower dexamethasone values, the predictive value of the DST for melancholia was not affected by the large variation in the bioavailability of dexamethasone.