Introduction: Hemodialysis patients (HD pts) with atrial fibrillation (AF) have increased risk of stroke. The HASBLED (Hypertension (HTN), Abnl Renal/Liver Function, Stroke, Bleeding Hx, Labile INR, Elderly, Drugs/Alcohol) risk score predicts bleeding in the general AF population. It is unknown whether the HASBLED score can be applied to HD pts who are at additional bleeding risk due to uremic platelet dysfunction and the regular use of heparin. Hypothesis: To address this question, we queried the United States Renal Data System (USRDS) for bleeding events in HD pts with AF, and correlated those events with a modified HASBLED (mHASBLED) score. Methods: All incident HD pts with AF from the USRDS for 2006-2010 were queried for major bleeding events and mHASBLED parameters using ICD-9 diagnosis codes and data from CMS form 2728. For mHASBLED, the HTN parameter was defined as "HTN as the cause of renal failure", and labile INR as > 16 INRs/yr, but all other parameters could be derived from the dataset. Logistic regression (LR) analysis was used to estimate the odds ratio (OR) for the mHASBLED score to predict major bleeding events. Results: 74,631 HD pts had AF, and 9.8% had a major bleeding event (GI bleeding and hemorrhagic stroke). By univariate analysis, those who bled were more likely to be elderly, have an underlying cause of renal disease due to HTN, prior bleeding event, hepatitis C, labile INR, and be on oral anticoagulants. By LR, variables with the greatest impact on bleeding were HTN as a cause of underlying renal disease, prior bleeding history, and labile INR (OR of 1.10, 2.20 and 2.24, respectively). The OR for bleeding events increased by 1.28 for each unit increase in mHASBLED. Older age, prior stroke, abnormal renal or liver function, and drug use had the least effect. Note that the lowest possible score in this cohort is 1, given that all patients had renal failure. Conclusions: In HD pts with AF, the mHASBLED predicts major bleeding events. The universal presence of renal disease, and the lack of specific clinical data from the USRDS may limit the clinical precision of a given score, however mHASBLED may remain a useful indicator of bleeding risk in this population.