OBJECTIVES: Warfarin had been the only effective oral anticoagulant (OAC) to reduce risk of ischemic stroke in patients with atrial fibrillation (AF) until recently new OACs (dabigatran, rivaroxaban and apixaban) became possible alternatives for AF patients. Cost-effectiveness of new OACs versus warfarin therapy in AF patients was examined. METHODS: A Markov model was designed to compare life-long economic and treatment outcomes of apixaban (5mg twice daily), dabigatran (150mg twice daily), rivaroxaban (20mg daily), and warfarin therapy at anticoagulation care with moderate control on anticoagulation (mean time in therapeutic range (TTR) 60.6%) in a hypothetical cohort of AF patients aged 65 years old with CHADS2 score 2. Model inputs were derived from literature, and outcome measure was incremental cost per quality-adjusted life-year (QALY) gained (ICER) from the healthcare provider’s perspective. Robustness of model was examined by sensitivity analysis. RESULTS:  In the base-case scenario, warfarin was the least costly alternative (USD82,768) and it gained lowest QALYs (9.571). ICERs of dabigatran and apixaban were USD38,382 and USD204,787, respectively. Rivaroxaban was more costly and gained less QALYs than dabigatran and apixaban. Using USD50,000 as the threshold of willingness-to-pay per QALY, dabigatran was the most cost-effective option. The most influential factors on the base-case results were the risk of bleeding and ischemic stroke of dabigatran comparing to warfarin therapy. In 10,000 Monte Carlo simulations, dabigatran was the most likely option to be cost-effective in 81% of time. CONCLUSIONS:  In the present model, the ICER of dabigatran was acceptable as the cost-effective alternative when compared to the other new OACs and anticoagulation service with moderate anticoagulation control.