OBJECTIVES: Population characteristics of patients with epilepsy remain poorly documented in France. Reimbursement databases may be useful to perform cost and epidemiological studies provided that patient diagnoses can be identified either directly, or indirectly through the use of antiepileptic drugs (AEDs) as markers. This study explored the possibility to use the French reimbursement database to determine the prevalence and direct cost of epilepsy. METHODS: The “EGB” reimbursement database is a 1/97 representative sample (500,000 individuals) of the population covered under the French General Scheme. Only a fraction of patients fully covered for epilepsy can be identified on a diagnosis basis. The rest of them can only be identified through their treatment by AEDs, but some are not specific to epilepsy (e.g. benzodiazepines). An algorithm was built to identify patients with epilepsy and calculate an estimation of the prevalence. In parallel, total medical expenses of patients were derived on the fully covered sub-population. RESULTS: Only patients treated with polytherapy (≥2 AEDs) could be identified in a relevant way by an algorithm based on drug use. The prevalence of epilepsy in this sub-group in 2009 was estimated between 1.83‰ and 2.79‰ (93,000 – 142,000 patients). A proportion of 70.1% to 71.6% were fully covered by insurance for their expenses, with epilepsy alone as a cause in only 27 to 33% of them. The most frequent comorbidities were psychiatric disorders and incapacitating stroke. The annual per capita expenses were in the range of 6601€- 6696€ for patients with polytherapy, inpatient care and drug costs represented about 50% and 27% respectively of overall expenses. The increase by 24.4% polytherapy patients mean costs as compared to monotherapy raised to 72% [IC 95: 44-106%] after adjustment on age, gender and presence of severe comorbidity. CONCLUSIONS: Polytherapy in epilepsy is associated with substantial higher direct costs.