The measurement of cortisol in saliva is becoming more widely accepted as a screening test for the diagnosis of hypercortisolism. Since 1986, cortisol measurement in saliva has been continuously used in our department. In this study we compared salivary cortisol profiles from proven Cushing's disease patients with profiles from healthy subjects and obese children. The purpose was to evaluate the predictive value of the method for the diagnosis of hypercortisolism and to define cut-off levels to exclude or identify hypercortisolism. Cortisol in saliva was measured in 150 Cushing's disease patients (30 children, 120 adults, ranging from age 4-70), 100 healthy subjects (55 children, 45 adults, ranging from age 6-60), and 31 children (age 7-15) with an age-related body-mass-index above the 90th percentile. Generally, five saliva samples were taken over the day at 6:00-8:00 a.m., 11:00-12:00 a.m., 4:00-6:00 p.m., 7:00-8:00 p.m., and 10:00 p.m. The samples were measured using a radioimmuno-assay (INCSTAR Corporation, Stillwater, Minnesota, USA). For healthy subjects, morning levels of cortisol in saliva between 3-19 microg/l were found. These levels dropped to levels in between <1-11 microg/l at 11:00-12:00 a.m., <1-6 microg/l at 4:00-6:00 p.m., <1-4.5 microg/l at 7:00-8:00 p.m., and <1-2.9 microg/l at 10:00 p.m. The measured values showed a correlation with age, height, and weight. In Cushing's disease patients, the circadian salivary cortisol rhythm was missing, compared to healthy subjects. There was no significant difference in salivary cortisol levels or circadian rhythm between healthy or obese children. We found a high sensitivity for the detection of hypercortisolism at the 10:00 p.m. salivary cortisol measurement. The following, age dependent cut-off levels for salivary cortisol at 10:00 p.m. were calculated for the exclusion of hypercortisolism. Age 6-10: 1.0 microg/l (specificity 100%, sensitivity 87.5%); age 11-15: 1.7 microg/l (specificity 100%, sensitivity 100%); age 16-20: 1.6 microg/l (specificity 100%, sensitivity 76.2%); age 21-60: 1.6 microg/l (specificity 100%, sensitivity 90.9%) [corrected] For the proof of Cushing's syndrome, the following age-dependent cut-off levels at 10:00 p.m. were found: age 6-10: 1.9 microg/l (specificity 100%, sensitivity 80%); age 11-15: 1.7 microg/l (specificity 100%, sensitivity 100%); age 16-20: 2.5 microg/l (specificity 100%, sensitivity 84.2%); age 21-60: 1.9 microg/l (specificity 100%, sensitivity 97.6 %) [corrected] The cortisol assessment in saliva is a sensitive and reliable method to discriminate normocortisolemic from hypercortisolemic patients. From our view, the major advantages of this method are the reliability, non-invasiveness, and use in ambulatory patients.