OBJECTIVE The high-dose dexamethasone (dex) suppression test of cortisol secretion (8 x 2 mg dex over two days or 8 mg overnight) is a mainstay in the differential diagnosis of Cushing's syndrome (CS). In some patients with pituitary Cushing's disease (CD), however, plasma cortisol is not suppressed to < 50% of control by 8 mg of dex. We therefore hypothesized that a higher dose of dex might produce more effective suppression of cortisol secretion in CD. METHODS We routinely tested the diagnostic efficacy of a very high dose of dex (32 mg, i.e. 4 x 8 mg in 24 hours) in comparison with the 8 mg overnight dex test in a population of patients with CD, in which an unusually high percentage was refractory to 8 mg dex. End points were the suppression of plasma cortisol, plasma ACTH and urinary free cortisol (UFC) to < 50% of control. Corticotrophin releasing hormone (human CRH) tests were also performed. RESULTS Eleven out of 26 (11/26) patients with CD (42%), among them six with pituitary macro-adenomas, failed to show suppression of plasma cortisol after 8 mg dex. Five out of 19 patients (26%) with CD failed to suppress after 32 mg dex. Only 3/19 (16%) failed to suppress UFC after 32 mg dex. In nonpituitary CS (n = 11), only one patient with macro-nodular adrenal hyperplasia showed significant suppression of plasma cortisol, but not UFC, after 32 mg dex. ACTH suppression after 8 or 32 mg dex was often less pronounced than that of cortisol and was of no diagnostic value. Cortisol stimulation by > or = 23% after hCRH injection differentiated 100% of patients with CD from other forms of CS. CONCLUSIONS In this series, the hCRH test was the most reliable test for the differential diagnosis of Cushing's syndrome. The 32 mg dexamethasone test with measurement of urinary free cortisol was clearly superior to the 8 mg test and to other aspects of the very high dose dexamethasone test. It can be recommended for 'non-suppressible' patients with ACTH-dependent Cushing's syndrome and can be performed on outpatients.