Summary. Background: Retinal vascular occlusions may constitute an independent risk factor for stroke in patients with atrial fibrillation. Methods: We performed a retrospective study on a nationwide cohort with atrial fibrillation from 1997 to 2008. The rate of stroke/systemic thromboembolism (TE)/transitory ischemic attack (TIA) was determined for atrial fibrillation patients with and without a history of retinal vascular occlusion. A Cox regression analysis, adjusted for risk factors and medications, was performed to determine the independent predictive value of retinal arterial or venous occlusion for the risk of ischemic stroke, TE or TIA in atrial fibrillation patients. Results: We included 87 202 patients with nonvalvular atrial fibrillation. At baseline, a history of retinal arterial occlusion was diagnosed in 224 patients (0.26%) and a history of retinal venous occlusion in 361 (0.41%). Patients without retinal occlusion had a rate of stroke/TE/ TIA of 4.52 (95% confidence interval [CI] 4.44-4.60). For patients with retinal arterial occlusion, the rate of stroke/ TE/TIA was 8.16 (95% CI 6.35-10.49) per 100 personyears, and for patients with retinal venous occlusion it was 7.28 (95% CI 5.93-8.94) per 100 person-years. In multivariate analysis, both retinal arterial occlusions (hazard ratio [HR] 1.39, 95% CI 1.08-1.79) and retinal venous occlusions (HR 1.26, 95% CI 1.02-1.54) were associated with an increased risk of future stroke/TE/TIA. Conclusions: A history of retinal arterial or retinal venous occlusion is associated with an increased risk of stroke/TE/TIA in patients with atrial fibrillation. Thus, prior retinal vascular occlusion may be considered as a previous thromboembolic event when evaluating stroke risk in patients with atrial fibrillation.