We studied plasma cortisol levels at 00:00 h and 08:00 h in 103 patients with Cushing's syndrome and 144 patients in whom this diagnosis had been excluded. These patients were hospitalized in our department from 1975 to 1996. Additionally, we measured these parameters in 20 healthy volunteers and in 5 patients with nonendocrine disease. Corresponding data of urinary free cortisol and low-dose dexamethasone suppression testing were included in the evaluation. Values (mean+/-SD) from patients with Cushing's syndrome: 510+/-232 nmol/l (range 165-1488) for plasma cortisol 00:00 h, 574+/-242 nmol/l (range 236-1612) for plasma cortisol 08:00 h, 991+/-885 nmol/24 h (range 154-4866) for urinary free cortisol and 479+/-304 nmol/l (range 34 - 1,393) for plasma cortisol after 1.5 mg dexamethasone. Values from the patients excluded from Cushing's syndrome: 99+/-76 nmol/l (range 5-371) for plasma cortisol 00:00 h, 393+/-136 nmol/l (range 119-812) for plasma cortisol 08:00 h, 126+/-84 nmol/24 h (range 30-485) for urinary free cortisol, and 64+/-82 nmol/l (range 5-395) for plasma cortisol after 1.5 mg dexamethasone. Values of the healthy volunteers respectively patients with non-endocrine disease: 59+/-30 nmol/l (range 25-130) respectively 127+/-80 nmol/l (range 62-265) for plasma cortisol 00:00 h and 388+/-144 nmol/l (range 157-651) respectively 498+/-113 nmol/l (range 302-581) for plasma cortisol 08:00 h. None of the Cushing patients exhibited a 00:00 h plasma cortisol below 140 nmol/l and only one had a urinary free cortisol below 200 nmol/24 h, whereas 4 were complete dexamethasone suppressors. The diagnostic value of these parameters was examined based on various cutoffs. We recommend determination of midnight plasma cortisol as an efficient and simple additional procedure for the diagnosis of Cushing's syndrome. The sensitivity and specificity of this procedure is similar to urinary free cortisol and slightly above the low-dose dexamethasone suppression testing in our hospitalized patients.