We assessed the value of midnight salivary cortisol for the initial diagnosis of Cushing's syndrome. Sixty-three patients with various causes of Cushing's syndrome (37 with Cushing's disease, 17 with adrenal Cushing's syndrome, and nine with ectopic ACTH syndrome) and 54 control subjects with simple obesity were studied. All patients with Cushing's syndrome excreted more than 90 microg urinary free cortisol (UFC)/d (248 nmol/d), and all controls excreted less than 90 microg/d UFC. All patients with Cushing's syndrome had a midnight salivary cortisol concentration above 2.0 ng/ml (5.52 nmol/liter), whereas only three controls did so [2.0 ng/ml (5.52 nmol/liter); 2.05 ng/ml (5.66 nmol/liter); and 3.6 ng/ml (9.96 nmol/liter)]. This cut-off provides a sensitivity of 100% and a specificity of 96%. In patients with Cushing's syndrome, midnight salivary cortisol concentrations were correlated with UFC collected over the same period of time (0800-0800 h). Salivary cortisol measurements taken every 4 h showed a typical lack of circadian variation. The daily measurement of midnight salivary cortisol concentrations for 2 wk or more in five other out-patients (with obvious Cushing's disease, subclinical adrenal Cushing's syndrome, suspected Cushing's syndrome, pituitary incidentaloma, and prolactinoma) demonstrated the clinical utility of this factor. Measuring midnight salivary cortisol is an easy and noninvasive means of diagnosing hypercortisolism. Its diagnostic accuracy is identical to, if not better than, that of previously described gold standards.