OBJECTIVES: To evaluate the cost-effectiveness of apixaban in the prevention of stroke in patients with non-valvular atrial fibrillation (NVAF) from a French payer perspective. METHODS: A Markov model was developed in accordance to the new French guidelines of the Commission for Economic Evaluation and Public Health (CEESP) that also recommends applying efficiency frontier. A hypothetical cohort of patients with NVAF, 70 years old eligible for stroke preventive treatment (CHADS2≥1) was simulated over lifetime. Clinical events modeled included strokes, systemic embolism, intracranial hemorrhage, other major bleeds, clinically relevant non-major bleeds and myocardial infarction. Treatment efficacy and bleeding data was obtained from ARISTOTLE. Efficacy data are from published indirect comparisons. Acute medical costs of stroke and hemorrhage were obtained from a dedicated analysis of the French national hospitalization database (PMSI). Long-term medical costs and utility data were derived from the literature. Univariate and probabilistic sensitivity analyses (PSA) were also performed to assess the robustness of the model projections. RESULTS: Three strategies were strictly dominated: aspirin, dabigatran 110mg, and dabigatran in sequential dosages. Two others were extendidly dominated: rivaroxaban and dabigatran 150mg. The efficiency frontier was constructed by connecting the remaining strategies: warfarin and apixaban. Apixaban resulted in an incremental additional cost of 2,870€ and an incremental QALY of 0.189, corresponding to an ICER of 15,157€/QALY versus warfarin. The PSA indicated that the probabilities for apixaban being cost-effective versus warfarin were respectively 80% and 90% at the informal willing-to-pay thresholds of 30,000€ and 50,000€. Sensitivity analysis identified stroke risk with apixaban and intracranial hemorrhage risk with warfarin as the variables most influencing the results. CONCLUSIONS: Apixaban may be the most economically efficient alternative to warfarin in NVAF patients eligible for stroke prevention in France. All other strategies are dominated. However, uncertainty surrounding these results should be investigated through real life data.