OBJECTIVE: Patients with medically intractable epilepsy are potential candidates for surgery if the epileptogenic tissue is localized and resectable. Surgical therapy can eliminate seizures but is very expensive. We followed a prospective adult cohort of intractable epileptic patients in order to perform a cost-effectiveness analysis. POPULATION AND METHODS: Adult patients with a suspected partial medically intractable and operable epilepsy were eligible for evaluation, explorations and/or surgery. Clinical and economical data were collected at the inclusion and every 6 months over at least two years. Two patient groups were analyzed: some underwent a surgery, others did not. Clinical data were compared between both groups. As the data collection was not yet complete, we compared the surgery to a continuation of the preoperative medical management in a cost-effectiveness analysis. Direct medical and nonmedical costs were evaluated according to a societal perspective. The effectiveness was defined as one year without seizure. We assessed the incremental cost-effectiveness ratio (ICER) for the first two years after the surgery. We also modeled long-term costs and effectiveness and extrapolated the results over the patients' lifetime with a Markov model. We computed the ICER and performed a sensitivity analysis. Indirect costs were measured in physical units and intangible costs were assessed with quality-of-life measures (QOLIE-31, SEALS). Data were compared before and after surgery. RESULTS: Among the 286 patients included, 119 did not enter in the analysis: 7 were not eligible, 44 not operable, 31 did not present a follow-up, 37 still underwent exams. Finally, 89 underwent a surgical treatment, and 78 were medically treated. Disease was more severe in surgical patients than in medical patients: seizures frequency, depressive disorders and cognitive impairment were greater. One year after the surgery, 83% patients were seizure free. During the year before inclusion and the year after surgery, direct costs were mainly due to hospitalization. During the second year after surgery, the cost of antiepileptic drugs predominated. One additional year without seizure costs 23 531 euro one year after surgery and 9533 euro two years after surgery. In a long-term perspective, the surgery became cost-effective between 7 and 8 years after the surgery. CONCLUSION: Surgical therapy is a cost-effective treatment in a middle-term even without indirect costs consideration.