We measured total and free plasma cortisol, 24-hour urinary cortisol excretion, and corticosteroid-binding globulin in 21 normal subjects, 25 patients with depressive illness, and 6 patients with Cushing’s disease. Patients with depression had mean 24-hour plasma (8.3 ± 2.7 µg/dl) and urinary (36 ± 33.55 µg/g creatinine) cortisol levels that did not differ from those of normal subjects (6.6 ± 1.7 µg/dl; 24.6 ± 15.4 µg/g creatinine), but were significantly lower than those of patients with Cushing’s disease (14.4 ± 2.4 µg/dl; 215 ± 101 µg/g creatinine). Not all patients with depression had hypercortisolemia, and the 1 -mg dexamethasone suppression test identified some of those with adrenal hyperfunction. 17 of 25 patients had normal 8 a.m. and/or 4 p.m. plasma cortisol after dexamethasone (suppressors), while 8 patients had values greater than 5 µg/dl (nonsuppressor). Suppressors had normal total 24-hour plasma and urinary cortisol, while nonsuppressors had levels that were in the range seen in Cushing’s disease. Patients with depression showed the expected circadian variation in total and free cortisol, but nonsuppressors had elevated levels in evening and early a.m. hours when levels in normal subjects were low. Patients with Cushing’s disease had elevated levels throughout the day. The mean binding capacity of corticosteroid-binding globulin was not different in normal and depressed subjects (23.9 ± 3.2 vs. 22.2 ± 3.4 µg/dl), but was significantly decreased in patients with Cushing’s disease (15.7 ± 3.5 µg/dl). Although total cortisol levels were similar, nonsuppressors had significantly lower mean 24-hour plasma free cortisol (1.01 ± 0.27 µg/dl) than patients with Cushing’s disease (2.4 ± 0.54 µg/dl). The modest elevation of free cortisol and its normal circadian pattern compared with patients with Cushing’s disease may account for the absence of physical findings of glucocorticoid excess in patients with depression who have adrenal hyperfunction.