OBJECTIVES: To evaluate the cost-effectiveness of rivaroxaban relative to warfarin, for the prevention of stroke in patients with AF. METHODS: A Markov model demonstrating the progression of AF patients from healthy state towards embolic and bleeding events and to death was adapted to the Turkish setting. The cycle length was set as three-months. The analysis was undertaken from a payer perspective. Event rates and treatment effects were derived from the ROCKET AF clinical trial. Utility values for events were based on international literature. Costs of each health state included year 2012 local costs of medications, monitoring and events (Mid-2012 USD currency rate was used). Incremental cost effectiveness ratios (ICER) per life year (LY) and quality-adjusted LY (QALY) gained were calculated. One-way sensitivity analyses were conducted to test the robustness of the model. The time horizon was life time period. Discount rate was set at 3.5% for economic and clinical inputs. Willingness-to-pay (WTP) threshold was set as twice the local gross domestic product per capita (US$20,888).  RESULTS: The total cost of rivaroxaban-treated patients was US$1,065 higher compared to warfarin. Additional drug acquisition costs (US$3,083) caused by rivaroxaban were mainly offset by reduced monitoring (US$1,902) and event costs (US$116). Moreover, rivaroxaban was associated with increments of 0.065LYs and 0.055QALY leading to an ICER of US$16,362/LY gained and US$19,500/QALY gained. Sensitivity analyses showed that the cost-effectiveness results are fairly insensitive to most inputs. The changes in the cost of embolic and hemorrhagic events had almost no effect on ICER values. The ICER value was sensitive to the cost and frequency of monitoring in warfarin patients. CONCLUSIONS: Rivaroxaban, with its cost-saving effect on monitoring,  reducing the frequencies of clinical events, improvement in LYs and QALYs, and ICER values below WTP threshold, is suggested to be a cost-effective alternative for the prevention of stroke in AF.