Transsphenoidal microsurgery is the standard treatment for patients with Cushing's disease. However, there is general lack of agreement regarding the definition of cure. We studied 58 patients with corrected hypercortisolism after transsphenoidal surgery for Cushing's disease. Plasma and urinary cortisol levels were measured after surgery. After the postsurgical hypocortisolism stage (or periodically in patients without hypocortisolism), urinary free cortisol, plasma cortisol at 0800 h and 2300 h, morning cortisol after 1 mg dexamethasone, and cortisol response to insulin-induced hypoglycemia were performed. Patients were classified in 3 groups: group I, patients with transient hypocortisolism and normal hypothalamus-pituitary-adrenal axis afterwards; group II, patients with transient hypocortisolism and abnormalities in the circadian rhythm or the stress response afterwards; and group III, patients without postoperative hypocortisolism. Thirty-three patients were included in group I, 8 in group II, and 17 in group III. Groups I and II were similar in postsurgical plasma cortisol (46.9 +/- 30.3 vs. 60.7 +/- 38.6 nM) and mean follow-up (69.8 vs. 68.8 months) but were significantly different in their recurrence rate (3.4% vs. 50%, P < 0.001). Patients in group III had normal postsurgical plasma and urinary cortisol but persistent abnormalities in circadian rhythm and stress response. After a mean follow-up of 39.1 months, their recurrence rate was similar to that of group II (64.7% vs. 50%). The complete normalization of the adrenocortical function, which is always preceded by postsurgical hypocortisolism, is associated with a very low recurrence risk and should be considered, in our opinion, the main criterion of surgical cure in Cushing's disease.