OBJECTIVES: Determine the most cost-effective, licensed, first-line anticoagulant for the prevention of ischemic stroke in patients with non-valvular atrial fibrillation (AF) in England and Wales from the perspective of the UK National Health Service. METHODS: We developed a cost-effectiveness model based on a review of previous models and expert clinical opinion. We compared warfarin (International Normalized Ratio  2-3) with novel oral anticoagulants (NOACs) apixaban (5mg bd), dabigatran (150mg bd), edoxaban (60mg od) and rivaroxaban (20mg od), over 30 years post treatment initiation. Parameters were informed by a systematic literature review and competing risks network meta-analysis for comparative efficacy and safety of the anticoagulants and baseline hazard of warfarin. Utilities and resource use were estimated from the literature. Our model estimated total costs, Quality Adjusted Life Years (QALYs), and incremental net benefit relative to warfarin. RESULTS: At a willingness-to-pay threshold of £20,000 per QALY, all NOACs have positive expected incremental net benefit compared to warfarin, suggesting they may be a cost effective use of NHS resources. Apixaban (5mg bd) has the highest expected incremental net benefit (£7533), followed by rivaroxaban (£6365), edoxaban (£5279) and dabigatran (£5279). Apixaban is the only NOAC for which the 95% credible interval around incremental net benefit is positive, suggesting a higher degree of certainty that Apixaban is cost-effective compared with warfarin than for the other NOACs. These conclusions also hold at the higher threshold of £30,000. CONCLUSIONS: At a willingness-to-pay threshold of £20,000 per QALY, all NOACs are cost-effective compared with warfarin. There is considerable uncertainty between the different NOACs, but apixaban (5mg bd) had the highest expected incremental net benefit and the highest probability (60%) of being most cost-effective first line anticoagulant for the prevention of stroke in AF, primarily due to lower rates of intracranial haemorrhage, other clinically relevant bleeding, and myocardial infarction.