We evaluated the accuracy of capillary whole blood international normalized ratio (INR) on the CoaguChek S (Roche Diagnostics, Indianapolis, IN), CoaguChek XS (Roche Diagnostics), and i-STAT 1 (i-STAT, East Windsor, NJ) point-of-care (POC) analyzers compared with venous plasma INRs determined by a reference laboratory method. Overall agreement between POC and laboratory plasma INR was very good, with median bias between capillary whole blood and laboratory plasma INRs varying from 0.0 to -0.2 INR units on all devices. More than 90% of results on the CoaguChek XS and i-STAT 1 and 88% of CoaguChek S results were within 0.4 INR units of the reference laboratory method. The CoaguChek XS and i-STAT 1 demonstrated greater accuracy than the CoaguChek S as measured by the number of results that differed by more than 0.5 INR units from the reference method. Median bias between CoaguChek S capillary whole blood and laboratory plasma INRs changed over time, demonstrating the need for ongoing quality assurance measures for POC INR programs.