The aim of this study was to determine whether preventing increases in plasma cortisol during antecedent hypoglycemia preserves autonomic nervous system counterregulatory responses during subsequent hypoglycemia. Experiments were carried out on 15 (8 male/7 female) healthy, overnight-fasted subjects and 8 (4 male/4 female) age- and weight-matched patients with primary adrenocortical failure. 5 d before a study, patients had their usual glucocorticoid therapy replaced with a continuous subcutaneous infusion of cortisol programmed to produce normal daily circadian levels. Both groups underwent identical 2-d experiments. On day 1, insulin was infused at a rate of 1.5 mU/kg per min, and 2-h clamped hypoglycemia (53+/-2 mg/dl) was obtained during the morning and afternoon. The next morning, subjects underwent an additional 2-h hypoglycemic (53+/-2 mg/ dl) hyperinsulinemic clamp. In controls, day 2 steady state epinephrine, norepinephrine, pancreatic polypeptide, glucagon, growth hormone, and muscle sympathetic nerve activity were significantly blunted (P < 0.01) compared with day 1 hypoglycemia. In marked contrast, when increases of plasma cortisol were prevented in the patient group, day 2 neuroendocrine, muscle sympathetic nerve activity, hypoglycemic symptoms, and metabolic counterregulatory responses were equivalent with day 1 results. We conclude that (a) prevention of increases of cortisol during antecedent hypoglycemia preserves many critical autonomic nervous system counterregulatory responses to subsequent hypoglycemia; (b) hypoglycemia-induced increases in plasma cortisol levels are a major mechanism responsible for causing subsequent hypoglycemic counterregulatory failure; and (c) our results suggest that other mechanisms, apart from cortisol, do not play a major role in causing hypoglycemia-associated autonomic failure.