The metyrapone test is used to test the hypothalamic-pituitary-adrenocortical axis. The present study aims to assess the diagnostic accuracy of combined stimulation of ACTH and compound-S (CMP-S). In addition, we analyzed the safety and practicability of this test as an outpatient procedure. A total of 327 metyrapone tests were analyzed retrospectively in 185 patients (mean age, 50.3 +/- 15.2 yr). One hundred thirteen patients had one test, and 72 patients had between 2 and 6 tests over 1-3 yr. Most patients suffered from pituitary adenomas (60 macroadenomas, 63 microadenomas) or other pituitary lesions (n = 29). Metyrapone (2 g) was given at 2400 h as an outpatient procedure. Blood samples for analysis of ACTH, CMP-S, and cortisol were taken at 0730 h. Stimulation of adrenal CMP-S and cortisol by pituitary ACTH demonstrated a dose-response curve with the shape of half a geometric parabola. CMP-S reached a plateau when ACTH rose above 175 ng/liter [r = 0.661, P < 0.0001 for ACTH <175 ng/liter; r = 0.083, P = not significant (NS) for ACTH >175 ng/liter], cortisol flattened at ACTH levels above 230 ng/liter (r = 0.633; P < 0.0001 for ACTH < 230 ng/liter; P = NS for ACTH >230 ng/liter). Alternatively, the sum of CMP-S plus cortisol also flattened when ACTH rose above 230 ng/liter (r = 0.696; P < 0.0001 for ACTH <230; P = NS for ACTH > 230 ng/liter). Receiver operating curve analysis defining a cut-off for ACTH at 150 ng/liter demonstrated a sensitivity of 47% and 67% at a cut-off level for CMP-S at 200 or 260 nmol/liter, respectively. The respective specificity was 82% and 68% for CMP-S. This compared with a sensitivity of 71% and specificity of 69% if the sum of CMP-S plus cortisol of 450 nmol/liter were used as cut-off. The response curve between CMP-S and ACTH implies a maximally stimulated adrenal cortex at circulating ACTH levels above 175 ng/liter. Single measurement of CMP-S using the cut-off at 200 nmol/liter, as suggested in the literature, yields a poor sensitivity of only 47% compared with ACTH. Despite the relatively high cross-reactivity of CMP-S in the cortisol assay, the sum of CMP-S and cortisol levels with a cut-off value of 450 nmol/liter yields a better diagnostic accuracy compared with CMP-S alone.