Plasma cortisol suppression after dexamethasone administration in 113 consecutively hospitalized patients (54 patients with primary depression, 41 with secondary depression and 18 non-depressed controls) showed early morning hypersecretion of cortisol and cortisol non-suppression after dexamethasone mainly in patients with primary depression. The sensitivity of the dexamethasone suppression test was 79 per cent in unipolar and bipolar primary depressed patients, specificity 79 per cent and diagnostic confidence of a positive test 82 per cent. Non-suppression to dexamethasone was observed in 81 per cent of patients with psychotic depression but only 37 per cent of those with nonpsychotic depression. Age, sex and severity of depression, menopausal status, or benzodiazepines did not influence results. There was no association between cortisol non-suppression and any genetic subgroup of affective disorder.