Measurement of late-night and/or midnight salivary cortisol currently used in US and European countries is a simple and convenient screening test for the initial diagnosis of Cushing's syndrome (CS). Unfortunately, this test has not been widely used in Japan. The purpose of this study was to evaluate the usefulness of the measurement of late-night salivary cortisol as a screening test for the diagnosis of CS in Japan. We studied 27 patients with various causes of CS, consisting of ACTH-dependent Cushing's disease [5] and ectopic ACTH syndrome [4] and ACTH-independent adrenal CS [11] and subclinical CS [7]. Eleven patients with type 2 diabetes and obesity and 16 normal subjects served as control group. Saliva samples were collected at late-night (23:00) in a commercially available device and assayed for cortisol by radioimmunoassay. There were highly significant correlations (P<0.0001) between late-night serum and salivary cortisol levels in normal subjects (r = 0.861) and in patients with CS (r = 0.788). Late-night salivary cortisol levels in CS patients (0.975 ± 1.56 μg/dl) were significantly higher than those in normal subjects (0.124 ± 0.031 μg/dl) and in obese diabetic patients (0.146 ± 0.043 μg/dl), respectively. Twenty-five out of 27 CS patients had late-night salivary cortisol concentrations greater than 0.21 μg/dl, whereas those in control group were less than 0.2 μg/dl. Receiver operating characteristic curve (ROC) analysis showed that the cut-off point of 0.21 μg/dl provides a sensitivity of 93% and a specificity of 100%. Therefore, it is concluded that the measurement of late-night salivary cortisol is an easy and reliable noninvasive screening test for the initial diagnosis of CS, especially useful for large high-risk populations, such as diabetes and obesity.