A number of factors contribute to interlaboratory variation of international normalized ratio (INR) results. This variability can lead to differences in clinical decision-making. This prospective cohort study evaluated the accuracy, precision, and clinical decision-making variability of two point-of-care (POC) devices (CoaguChek XS and the Coag-Sense) to a clinical laboratory method of INR measurement as the ‘reference’ and by using the laboratory's accuracy standard (±25% of the reference INR). Study subjects included adults taking warfarin undergoing a POC INR with each device and a laboratory INR on the same day and evaluated differences in clinical decision-making using a warfarin nomogram. A correction factor was derived for each POC method to estimate the corresponding laboratory INR for POC INRs more than 3.0. A total of 100 patient encounters (76 unique patients) included INR testing by all three methods. Both POC devices demonstrated excellent precision across the entire range (r = 0.98), but poor accuracy relative to the laboratory standard for POC INRs greater than 3.0 (accuracy CoaguChek 19% and Coag-Sense 35%). A correction factor applied to POC INRs greater than 3.0 improved accuracy to 99% for both devices. Applying the correction factor also significantly reduced differences in clinical decision-making (49–28% for CoaguChek and 56–40% for Coag-Sense, P < 0.001). Both POC devices demonstrated poor accuracy for INRs over 3.0. A device-specific, institution-specific correction factor applied to POC INRs greater than 3.0 can provide better accuracy of INR measurement and can significantly reduce variability in warfarin dosing decisions. More study is needed to assess impact on clinical adverse events.