Clinical features such as weight gain, depression, hypertension, and menstrual irregularities, although common in the general population, may raise the possibility of Cushing's syndrome. Up to 30% of urine cortisol and dexamethasone suppression screening tests may return an incorrect result, suggesting that better tests are needed. This study evaluated the utility of nighttime salivary cortisol measurement as a screening test for Cushing's syndrome. We evaluated 139 inpatients and 4 outpatients with possible Cushing's syndrome, 16 inpatients and 7 outpatients with other nonadrenal disorders, and 34 healthy outpatients. Using cut points that excluded all subjects without Cushing's syndrome, we compared the sensitivity for the detection of Cushing's syndrome of nighttime salivary cortisol levels (2330 and 2400 h for inpatients and bedtime for outpatients), simultaneous inpatient serum cortisol levels, and urine glucocorticoid excretion. An assay- specific inpatient 2400-h salivary cortisol or an outpatient bedtime salivary cortisol greater than 550 ng/dl (15.2 nmol/liter) identified 93% of patients with Cushing's syndrome (confidence interval, 89-98%) and excluded all individuals without the disorder. Salivary cortisol measurements worked as well as plasma measurements and better than urine glucocorticoid excretion. We concluded that bedtime salivary cortisol measurement is a practical and accurate screening test for the diagnosis of Cushing's syndrome.