Context: Although transsphenoidal surgery remains the first-line treatment in Cushing's disease (CD), recurrence is observed in about 20% of cases. Adjunctive treatments each have specific drawbacks. Despite its inhibitory effects on steroidogenesis, the antifungal drug ketoconazole was only evaluated in series with few patients and/or short-term follow-up. Objective: Analysis of long-term hormonal effects and tolerance of ketoconazole in CD. Design: A total of 38 patients were retrospectively studied with a mean follow-up of 23 months (6-72). Setting: All patients were treated at the same Department of Endocrinology in Marseille, France. Patients: The 38 patients with CD, of whom 17 had previous transsphenoidal surgery. Intervention: Ketoconazole was begun at 200-400 mg/day and titrated up to 1200 mg/day until biochemical remission. Main outcome measures: Patients were considered controlled if 24-h urinary free cortisol was normalized. Results: Five patients stopped ketoconazole during the first week because of clinical or biological intolerance. On an intention to treat basis, 45% of the patients were controlled as were 51% of those treated long term. Initial hormonal levels were not statistically different between patients controlled or uncontrolled. Ketoconazole was similarly efficacious as a primary or postoperative treatment. Among 15 patients without visible adenoma at initial evaluation, subsequent follow-up allowed identification of the lesion in five cases. No adrenal insufficiency was observed. Adverse effects were rare in patients treated long term. Conclusions: Ketoconazole is a safe and efficacious treatment in CD, particularly in patients for whom surgery is contraindicated, or delayed because of the absence of image of adenoma on magnetic resonance imaging.European Journal of Endocrinology 158 91-99