Unstimulated early morning cortisol has been suggested as a first line parameter to assess adrenal function in patients with suspected secondary adrenal insufficiency. The measurement of basal salivary cortisol (BSaC) instead of basal serum cortisol (BSeC) offers some advantages, such as painless sampling and the determination of the free hormone. The objective of this study was to evaluate the diagnostic value of BSeC and BSaC in comparison to the insulin tolerance test (ITT). Seventy-seven patients with hypothalamic-pituitary disease and 184 healthy controls were enrolled. ITT were performed in patients, and BSeC as well as BSaC levels were measured in patients and controls. Upper and lower thresholds (with >or=95% specificity either for adrenal sufficiency or adrenal insufficiency) were calculated by ROC analysis both for BSeC and BSaC. The ITT identified 41 patients as adrenal insufficient and 36 patients as adrenal sufficient. Upper and lower cutoffs were 470 and 103 nmol/l for BSeC, and 21.1 and 5.0 nmol/l for BSaC, respectively. Thereby, basal cortisol allowed a highly specific diagnosis (i.e., similar to the ITT result) in either 23% (BSeC) or 27% (BSaC) of patients. We suggest the determination of unstimulated early morning cortisol as first-line screening method for the diagnosis of secondary adrenal insufficiency. If upper and lower cutoffs are used, dynamic testing could be obviated in about one fourth of cases. Due to its easy and painless collection BSaC may be preferable to BSeC.