To estimate the cost-effectiveness of stroke prevention in patients with nonvalvular atrial fibrillation by using novel oral anticoagulants apixaban 5 mg, dabigatran 150 mg, and rivaroxaban 20 mg compared with warfarin. A Markov decision-analysis model was constructed using data from clinical trials to evaluate lifetime costs and quality-adjusted life-years of novel oral anticoagulants compared with warfarin. The modeled population was a hypothetical cohort of 70-year-old patients with nonvalvular atrial fibrillation, increased risk for stroke (CHADS2 ≥ 1), renal creatinine clearance ≥ 50 mL/min, and no previous contraindications to anticoagulation. The willingness-to-pay threshold was $50 000/quality-adjusted life-years gained. In the base case, warfarin had the lowest cost of $77 813 (SD, $2223), followed by rivaroxaban 20 mg ($78 738 ± $1852), dabigatran 150 mg ($82 719 ± $1959), and apixaban 5 mg ($85 326 ± $1512). Apixaban 5 mg had the highest quality-adjusted life-years estimate at 8.47 (SD, 0.06), followed by dabigatran 150 mg (8.41 ± 0.07), rivaroxaban 20 mg (8.26 ± 0.06), and warfarin (7.97 ± 0.04). In a Monte Carlo probabilistic sensitivity analysis, apixaban 5 mg, dabigatran 150 mg, rivaroxaban 20 mg, and warfarin were cost-effective in 45.1%, 40%, 14.9%, 0% of the simulations, respectively. In patients with nonvalvular atrial fibrillation and an increased risk of stroke prophylaxis, apixaban 5 mg, dabigatran 150 mg, and rivaroxaban 20 mg were all cost-effective alternatives to warfarin. The cost-effectiveness of novel oral anticoagulantss was dependent on therapy pricing in the United States and neurological events associated with rivaroxaban 20 mg.