Introduction:Atrial fibrillation (AF) is associated with an increased risk of stroke. Despite a large body of evidence linking malignancy and thrombosis, malignancy is not part of the scoring system.Hypothesis:Malignancy is an independent risk factor for the development of thromboembolic stroke in patients with AF.Methods:SEER database was utilized to identify patients with lung, colon, breast, and prostate cancers with AF and no prior diagnosis of stroke. These patients were compared to controls within the dataset. The primary endpoint was ischemic stroke. Rates of stroke per 100 person-years were calculated for each group. Cox regression modeling and a nested model comparing CHA2DSsVASc score (Model 1) with a complete model including cancer diagnosis (Model 2) were performed. Models were compared using Akaike Information Criterion (AKI) and Net Reclassification Index (NRI). A propensity matched cohort with equivalent CHA2DS2VASc scores determining stroke-free survival was also performed.Results:A total of 101,185 patients were included in the analysis, with 48,242 in the Cancer Group and 52,943 in the Non-cancer Group. Stroke rate per 100 person-years was significantly higher in the Cancer Group (2.962) compared with the Non-cancer Group (2.377) (IRR 1·25, 95% CI [1·18-1·31]). The CHA2DS2VASc model (Model 1) was compared against a model including cancer (Model 2) showing an improved predictability as assessed by both NRI (12·2, 95% CI [10·7-13·7, p<0.01) and AKI. Cox regression analysis calculated a hazard ratio of 1.085 for Cancer (95% CI [1·029-1·44], p = 0·0027), which was comparable to age > 75, female sex, and diabetes. Propensity matched Kaplan-Meier curve demonstrated decreased probability of stroke-free survival in the Cancer Group (p=0·00045).Conclusion:Cancers increase the risk of stroke in patients with AF. Consideration should be given for the addition of malignancy to the clinical scoring systems.