OBJECTIVES: Topiramate as add-on treatment is effective in patients with refractory epilepsy. Since there has been less research aimed at evaluating its cost-effectiveness, we conducted a retrospective lifetime cost-utility analysis. METHODS: Effectiveness data were derived from a pivotal placebo-controlled clinical trial, while utility and cost data were retrieved from published information. Our pharmacoeconomic model was similar to that employed in a previous study on lamotrigine (Eur J Clin Pharmacol 1998;53:421), but used UK cost-of-illness data. In the first analysis, aggregated data on effectiveness were used (clinical endpoint=number of patients with 50% or more reduction in seizure frequency) allowing a direct comparison between lamotrigine and topiramate. Our second analysis was based on individual patient data from the topiramate trial (clinical endpoint=individual values of reduction in seizure frequency). Both analyses were intention-to-treat. RESULTS: The analysis based on aggregated data showed that topiramate (200 mg/day) implies an incremental lifetime cost of £350,326 and a utility of 90 quality-adjusted life-years (QALYs) for every 100 patients. The same figures for lamotrigine (500 mg/day) were £1,324,295 and 58 QALYs, respectively (discounted values, yearly rate=3%, year of costing=2001). The cost-utility ratio was £3,893 per QALY gained for adjunctive topiramate as opposed to £22,833 for lamotrigine, both compared to add-on placebo. Sensitivity testing suggested a range of £3,129 to £4,870 and of £14,175 to £45,841 for topiramate and lamotrigine, respectively. Our analysis based on individual data showed that topiramate compared to placebo implies an incremental lifetime cost of £1,024,941 and a utility of 48 QALYs for every 100 patients. The cost utility ratio is £21,353 per QALY gained. CONCLUSIONS: Our results show that, in refractory epilepsy, adjunctive topiramate has a favourable pharmacoeconomic profile. In this analysis, using aggregated clinical information generated more optimistic values of cost effectiveness than using individual patient data.