OBJECTIVES: Assessment of point of care devices (POCs) that measure INR does not typically include the impact of measurement error on clinical decision-making. This study determined the effect of POC measurement error on warfarin dosing decisions, and how this effect varies over the INR scale. METHODS: For each patient, INR was simultaneously measured using two different POC devices and standard laboratory techniques. Clinicians blinded to INR measurement technique were asked to state a warfarin dosing decision for each INR value. The difference between each POC-derived INR and the laboratory standard measure were recorded for each patient, as were differences in warfarin dosing decisions based on the POC devices versus laboratory standard measures. RESULTS: A total of 202 patients on warfarin therapy were enrolled. Overall, the POC devices resulted in different warfarin dosing decisions compared to the laboratory standard in 114 of 404 instances (28%, 95% CI: 24-33%). Discordant dosing decisions resulting from POC and laboratory measures were most common when the laboratory INR was between 1.5-1.99 (40/92, 43%, 95% CI: 33-55%) and 3-3.49 (15/36, 42%, 95% CI: 26-59%). Relatively small amounts of measurement error from the POC devices resulted in different warfarin dosing decisions between the laboratory and POCs at specific intervals in the INR scale. For example, overestimation of INR by the POC by 0.2-0.39 INR units resulted in discordant warfarin dosing decisions in 44% of instances when the laboratory INR was between 1.5-1.99. CONCLUSIONS: Measurement error in POC devices leads to different warfarin doses in a high proportion of cases. A relatively small amount of measurement error can result in different warfarin dosing decisions if it occurs at specific intervals on the INR scale. Investigators and regulators should consider the varying effect of measurement error along the INR scale and its impact on clinical decision-making when evaluating POCs.