Despite unequivocal published evidence that patients with subnormal hypothalamic-pituitary-adrenal (HPA) function may respond normally to ACTH, such normal results are still considered reliable indications of unimpaired HPA function. This view was tested in four patients with clinical features suggesting corticotropin deficiency, in whom cosyntropin (0.25 mg, i.v.) raised serum cortisol above 560 nmol/L (20 micrograms/dL) at 1 h. All four patients had subnormal responses to metyrapone and excellent persistent improvement during subsequent glucocorticoid therapy. Serum cortisol concentrations 1 h after cosyntropin treatment in these patients closely resembled cortisol concentrations 1 h after uncomplicated cholecystectomy in six other patients. However, the rapid ACTH test in the patients with hypopituitarism failed to indicate whether more prolonged stimulation by ACTH or their endogenous stress would stimulate the normal continuing rise in serum cortisol, which reached 1358 +/- 170 nmol/L (+/- SE) 5 h after the incision in the cholecystectomized patients. As the three hypocorticotropic patients who were recognizably stressed had unstressed serum cortisol levels despite persistent adrenocortical reserve (shown by their ACTH responses) and recovered during glucocorticoid therapy, the ACTH test, if interpreted to indicate normal HPA function, would probably have had disastrous consequences. We conclude that a normal response to the rapid ACTH test can be dangerously misleading, particularly in incomplete ACTH deficiency states.