Differentiating the cause of Cushing's syndrome traditionally has depended upon measuring the response of 24-hour urine samples of cortisol or glucocorticoid metabolites to the high-dose (8 mg per day) dexamethasone test. The metyrapone test, however, is more convenient because it is a shorter test and requires the obtainment of serum samples, which can be collected more simply and more reliably than 24-hour urine samples. The usefulness of these two tests has not been adequately evaluated in a large series of patients with Cushing's syndrome. This study prospectively evaluated the accuracy of the dexamethasone and metyrapone tests in determining the cause of Cushing's syndrome in a series of 25 unselected patients. The diagnostic accuracy of these tests was calculated as follows: diagnostic accuracy = true positives and true negatives/study population X 100. Results of this study demonstrated that the metyrapone test was more accurate than the dexamethasone test in differentiating Cushing's disease from adrenocortical neoplasm (diagnostic accuracy, 100 percent versus 81 percent). All patients with Cushing's disease had a normal postmetyrapone 11-deoxycortisol concentration (greater than 10 micrograms/dl), while all patients with adrenocortical neoplasm had a suppressed 11-deoxycortisol concentration (less than 10 micrograms/dl). Thus, this study demonstrates that the metyrapone test is superior to the high-dose dexamethasone test in the differential diagnosis of Cushing's syndrome.