A normal plasma cortisol response to exogenous corticotropin has been advanced as a reliable indication of adequate hypothalamic-pituitary-adrenal function in patients suspected of having secondary adrenal failure. We have examined the validity of this diagnostic strategy in five patients who had undergone hypophysectomy and 27 patients recently treated with glucocorticoids. Eleven of the patients had normal adrenal responses to cosyntropin but had subnormal responses when the entire hypothalamic-pituitary-adrenal axis was examined using metyrapone; no patient who responded normally to metyrapone failed to respond to cosyntropin. Inducing hypoglycemia with insulin yielded results concordant with the results induced by metyrapone in four patients tested. A normal cortisol response to corticotropin alone should not be relied on to exclude secondary adrenal insufficiency. To do this it is necessary to demonstrate normal activity of the entire hypothalamic-pituitary-adrenal axis, which can be conveniently examined using metyrapone.