BACKGROUND: Anticoagulation management is difficult in chronic kidney disease, with frequent supratherapeutic international normalized ratios (INRs ≥ 4) increasing hemorrhagic risk. We evaluated whether the interaction of INR and lower estimated glomerular filtration rate (eGFR) increases hemorrhage risk and whether patients with lower eGFRs experience slower anticoagulation reversal. STUDY DESIGN: Prospective cohort study. SETTING & PARTICIPANTS: Warfarin pharmacogenetics cohort (1,273 long-term warfarin users); warfarin reversal cohort (74 warfarin users admitted with INRs ≥ 4). PREDICTOR: eGFR, INR as time-dependent covariate, and their interaction in the pharmacogenetics cohort; eGFR in the reversal cohort. OUTCOMES & MEASUREMENTS: In the pharmacogenetics cohort, hemorrhagic (serious, life-threatening, and fatal bleeding) risk was assessed using proportional hazards regression. In the reversal cohort, anticoagulation reversal was assessed from changes in INR, warfarin and metabolite concentrations, clotting factors (II, VII, IX, and X), and PIVKA-II (protein induced by vitamin K absence or antagonist II) levels at presentation and after reversal, using linear regression and path analysis. RESULTS: In the pharmacogenetics cohort, 454 (35.7%) had eGFRs < 60 mL/min/1.73 m(2). There were 137 hemorrhages in 119 patients over 1,802 person-years of follow-up (incidence rate, 7.6 [95% CI, 6.4-8.9]/100 person-years). Patients with lower eGFRs had a higher frequency of INR ≥ 4 (P<0.001). Risk of hemorrhage was affected significantly by eGFR-INR interaction. At INR<4, there was no difference in hemorrhage risk by eGFR (all P ≥ 0.4). At INR≥4, patients with eGFRs of 30 to 44 and < 30 mL/min/1.73 m(2) had 2.2-fold (95% CI, 0.8-6.1; P=0.1) and 5.8-fold (95% CI, 2.9-11.4; P<0.001) higher hemorrhage risks, respectively, versus those with eGFRs ≥ 60 mL/min/1.73 m(2). In the reversal cohort, 35 (47%) had eGFRs < 45 mL/min/1.73 m(2). Patients with eGFRs < 45 mL/min/1.73 m(2) experienced slower anticoagulation reversal as assessed by INR (P=0.04) and PIVKA-II level (P=0.008) than those with eGFRs ≥ 45 mL/min/1.73 m(2). LIMITATIONS: Limited sample size in the reversal cohort, unavailability of antibiotic use and urine albumin data. CONCLUSIONS: Patients with lower eGFRs have differentially higher hemorrhage risk at INR ≥ 4. Moreover, because the INR reversal rate is slower, hemorrhage risk is prolonged.