OBJECTIVES: Adopting new medical therapies is a complex decision that must take into account many factors, including differences in efficacy, tolerability, safety, and costs. Long-term comparative trials, especially among newer antiepileptic drugs (AEDs), are lacking, therefore decision models are needed to guide treatment decisions. We aimed to develop an economic model of the treatment of refractory epilepsy in the UK, and to assess the cost-effectiveness of topiramate as adjunctive treatment in refractory epilepsy compared to other newer AEDs. METHODS: A Markov model was developed to combine data from published clinical trials, cost-of illness studies, epilepsy-related mortality surveys, and utility studies. The expected costs and utilities associated with possible treatment strategies (1st and 2nd line add-on treatments) for newly diagnosed epilepsy patients with partial seizures were calculated and compared. In those patients requiring a second-line add-on, it was assumed that the first-line add-on treatment was stopped. A probabilistic analysis was undertaken and the cost-effectiveness frontier mapped. RESULTS: First and second-line adjunctive treatment with topiramate followed by levetiracetam was the least costly add-on strategy, and this strategy had the highest probability of being cost-effective at currently accepted values of the ceiling ratio (£60,000/QALY), while adjunctive treatment with levetiracetam and lamotirigine are both more expensive and generate less QALYs than the other scenarios, therefore cannot be preferred. CONCLUSIONS: This model suggests that topiramate first-line adjunctive treatment followed by levetiracetam second-line (or vice versa) are cost-effective treatment strategies in patients with partial seizures refractory to other treatments.