We surveyed 271 laboratories participating in a quality assessment program to ascertain whether the use of a calibration curve for determining the international normalized ratio (INR) would improve interlaboratory accuracy and precision. Lyophilized warfarinized samples with INR values assigned through manual calibration against internationally assigned rabbit reference thromboplasts were assayed for prothrombin time. Calibration analysis on the results was performed by linear regression. In all but 1 sample, the mean INR value computed by the calibration method was closer to the "true" value than the mean for the conventional calculation method using the International Sensitivity Index (ISI); the ISI calculation consistently overestimated the true value. Interlaboratory variation decreased using the calibration method. Variation from reagent to reagent was greater than from instrument to instrument, but was reduced by the calibration method. The specificity of the ISI for instrument type did not seem to alter the findings. Use of in-house calibrators to verify the ISI improved precision but not necessarily accuracy. The formation of a stable calibration line is consistent over time, but further studies are required to confirm whether such calibration improves the accuracy and precision of INR determination in practice.