OBJECTIVES: To compare costs and effectiveness of dabigatran etexilate (DAB) versus warfarin (WAR) in patients with Non-Valvular Atrial Fibrillation (NVAF) from a private and public health care system perspective in Brazil. METHODS: A Markov model was built to compare DAB versus WAR to derive the incremental cost effectiveness ratio (ICER) of DAB, based on the international literature and a modified Delphi panel with Brazilian experts (local clinical practice pattern on the management of NVAF patients). The model estimated the number of ischaemic and haemorrhagic strokes, systemic embolisms, intracranial hemorrhages, transient ischaemic attacks, extracranial hemorrhages, minor bleeds and acute myocardial infarctions associated with the respective treatments. To each clinical event costs, disabilities and/or reduction in quality of life, and risk of death were assigned. Only direct medical costs were considered and a discount rate of 5% was assumed, according to Brazilian HTA guidelines. A probabilistic sensitivity analysis was designed to assess uncertainty. RESULTS: Under both, the private and public perspective, DAB was associated with additional 0.30 life years gained (LY), additional 0.35 QALYs and demonstrated a lower incidence of intracranial events versus WAR, resulting in lower event costs and follow-up costs. The ICER for DAB versus WAR was R$ 39,740/LY and R$ 34,867/QALY from the public and R$ 25,252.48/LY and R$ 22,160.20/QALY from the private perspective. Sensitivity analyses confirmed the cost-effectiveness of DAB. CONCLUSIONS: Findings suggest that DAB can be cost-effective for stroke prevention when used instead of WAR in NVAF patients in Brazil, given that the ICERS were below the threshold of other technologies reimbursed.