The dexamethasone (DXM) test has been widely used for diagnosing Cushing's disease (CD). The purpose of this paper is to review its diagnostic merit based on calculation of data extracted from earlier publications. Studies presenting individual values for patients with CD and normal subjects were identified through PubMed searches and references in pertinent studies. Calculation of the retrieved data demonstrated huge variation in the relative suppressibility, negative suppression being common. Furthermore, in almost each study retrieved, the pre and post DXM values were closely correlated. Finally, the generally accepted view that DXM causes less suppression in Cushing's disease than in euadrenal controls appears unfounded. A central issue in the definition of so-called "pseudo-Cushing's states" is failure to suppress cortisol secretion with DXM. From analysis of the literature it appears quite possible that this does not reflect a specific endocrine deficit, but a physiological "stress" reaction. The above issues question the diagnostic value of the test, in particular in clinically and biochemically borderline cases.