Objective: Several tests have been proposed to diagnose patients with Cushing's syndrome (CS). The aims of the study were: i) to evaluate the performance of salivary cortisol (SC) in hypercortisolism and ii) to compare SC with serum cortisol (SeC) and urinary cortisol. Design and patients: This was a diagnostic study. Twenty-seven patients with untreated Cushing's disease (CD untr), 21 women consuming oral contraceptive pill (OCP), 18 pregnant women, and 89 healthy subjects (controls) were enrolled. Methods: SC and SeC at baseline and after the low-dose dexamethasone suppression test (LDDST) and urinary free cortisol (UFC) were measured. Results: Midnight SC had a sensitivity of 100% in the CD untr group and a specificity of 97.7% in the controls. Specificity remained high (95.2%) in women taking OCP, while in pregnant women, it decreased to 83.3%. SC after the LDDST showed a sensitivity of 96.3% in the CD untr group; specificity was 97.7% in the controls and 90.5% in OCP women. Midnight SeC had a sensitivity of 100% in the CD untr group. SeC after the LDDST had a sensitivity of 100% in the CD untr group while specificity was 97.7% in the controls and 61.9% in women taking OCP. For UFC, sensitivity was 92.6% in the CD untr group while specificity was 97.7% in the controls and 100% in the OCP group. Conclusions: SC is a reliable parameter for the diagnosis of severe hypercortisolism, with high sensitivity and specificity. In women during pregnancy or taking OCP, the measurement of SC, identifying the free fraction, could be helpful to exclude CS.