OBJECTIVES: U.S. guidelines recommend anticoagulating patients with atrial fibrillation (AF) and CHADS-VASc score of 2 or greater. This recommendation relies on stoke risk from the SPORTIF trials; a score of 2 represents 2.2% annual incidence of stroke. New evidence from multiple AF cohorts indicates the risk of stroke off anticoagulants may vary considerably. This variation in stroke risk may substantially impact the net benefit of anticoagulation. We used stroke rates from SPORTIF and the larger U.S. community-based ATRIA cohort to address this question. For given CHADS-VASc scores, stroke rates in ATRIA are, on average, 2% lower than in SPORTIF. METHODS: We conducted a simulation study of 33,436 patients with incident AF within the ATRIA-CVRN Kaiser California cohort. We determined the gain in quality-adjusted life years (QALYs) associated with anticoagulation compared with no thromboprophylaxis, using a previously published Markov state transition decision model incorporating patient-specific stroke and hemorrhage risk factors. We performed two simulations using stoke rates reported in (1) SPORTIF and (2) ATRIA. We report the number of patients expected to benefit from anticoagulation (gain ≥ 0.1 QALYs). We also determined population and mean gain in QALYs/patient through guideline-based anticoagulation. RESULTS: Simulations using SPORTIF rates lead to expected net benefit for 74.6% (24,960) of the cohort while simulations using ATRIA rates lead to expected net benefit for 26.4% (8,817). Using SPORTIF rates, guideline-based anticoagulation (i.e., CHADS-VASc 2+), would lead to a gain of 12,716 QALYs (mean, 0.47/per patient receiving anticoagulation therapy). By contrast, using ATRIA rates, guideline-based anticoagulation would lead to a gain of 966 QALYs (mean, 0.04/patient receiving anticoagulant therapy). CONCLUSIONS: The net benefit of anticoagulation for AF is highly sensitive to variation in baseline stroke rates. Future work is needed to reconcile the observed variation in baseline stroke risk so that the benefit of anticoagulation can be accurately determined.