Adrenal insufficiency (AI) induced by glucocorticoids was first described more than 50 years ago in patients undergoing surgical stress. Although considered the most frequent cause of AI, the true incidence of this complication of glucocorticoid treatment remains unknown. No factors are known to predict AI after glucocorticoid treatment. In particular, neither the dose nor the duration of treatment seems predictive. The minimum dose of cortisol necessary for the body to cope with medical or surgical stress is unknown. The adrenocorticotropin test is often used during corticosteroid withdrawal because it is well correlated with adrenal response to surgical stress, but not with clinical events. Studies over the past 15 years have shown that the perioperative risk of AI has been overestimated and that hydrocortisone doses should be decreased. A prospective study of patients after steroid withdrawal is the only means of assessing the true incidence of this complication to propose a rational strategy to prevent it.