OBJECTIVES: To assess the cost-effectiveness of rivaroxaban versus acenocumarol and a real-world prescription pattern (RWP), for prevention of stroke in atrial fibrillation (AF) patients in the Spanish NHS setting. METHODS: A Markov model was developed with health and treatment states describing the management and consequences of AF. Patients enter the model with stable non-valvular AF are treated with oral anti-coagulants or a RWP (53.7% patients treated with acenocumarol, 32.6% with acetylsalicylic acid and 13.7% untreated, data extracted from a Longitudinal Patient Database) and progress between states according to transition probabilities derived from the ROCKET-AF study (safety-on-treatment dataset), observational data and a network meta-analysis. Costs (drug, monitoring and events costs) and outcomes were assigned to each state, and used to estimate costs per quality-adjusted-life-year gained (QALY). A disutility for acenocumarol was included. The cycle length was 3 months and time horizon was set to describe the lifetime of the patients (3% annual discounting rate for costs and benefits). A sub-population analysis compared rivaroxaban to acenocumarol in patients with poor INR control (patients who spend less than 60% of time in therapeutic range). Probabilistic sensitivity analysis (PSA) was conducted to test the robustness of results. RESULTS: In base-case, drug acquisition costs were higher in rivaroxaban group (€5,117 incremental cost), while cost savings were estimated for monitoring and events avoided compared to acenocumarol group (-€2,091 and -€522, respectively) and RWP (-€1,385 and -€2,366, respectively) with a higher effectiveness (0.20 and 0.51 incremental QALYs, respectively). ICERs were €11,274/QALY vs. acenocumarol and €3,986/QALY vs. RWP. In poor INR controlled group the ICER was €2,144/QALY. PSA indicated a high probability of rivaroxaban being cost-effective at a threshold of €30,000/QALY. CONCLUSIONS: In the Spanish setting rivaroxaban is a cost-effective treatment option compared to VKA in the total AF population and particularly in poorly controlled VKA patients.