OBJECTIVES: Discontinuation rates for vitamin K antagonists (VKAs) are high in patients with atrial fibrillation (AF). The aim of the current study was to assess the impact of VKA persistence compared to that of rivaroxaban, a new oral anticoagulant (NOAC) taken once daily and requiring no monitoring, on the incidence of ischemic strokes in patients with AF. METHODS: Data relating to persistence rates of warfarin and rivaroxaban over 18 months of patient follow-up was derived from the literature. A model was developed synthesising these data with published ischaemic stroke rates and the effect of switching to an alternative treatment. The model assumed an annual risk of ischaemic stroke of 1.65% and 5.00% for patients receiving warfarin or no treatment, respectively. To obtain the risk of stroke for patients receiving rivaroxaban, a relative risk of 0.94 was applied to the warfarin risk giving an annual probability of 1.55%. Publications suggest that persistence after 180 days is 44% for warfarin and 78% for rivaroxaban. Sensitivity analyses on persistence rates of rivaroxaban were conducted to account for uncertainty. RESULTS: Patients starting treatment with a VKA stayed on treatment for an average of 226 days. In contrast, patients starting on rivaroxaban stayed on treatment for approximately 412 days. The total ischaemic stroke risk at 18 months was estimated to be 5.15% and 3.48–4.07% for patients initiating therapy with warfarin or rivaroxaban, respectively. For a hypothetical cohort of 10,000 patients with AF the sensitivity analyses showed that the greater persistence with rivaroxaban would translate to 108–167 strokes avoided. CONCLUSIONS: Our model indicates that starting patients on rivaroxaban may decrease the number of total ischaemic strokes relative to warfarin treatment. Moreover, this decrease may correspondingly reduce the burden of stroke-related costs borne by healthcare systems.