Background: Few studies have examined the accuracy of patient-reported international normalized (INR) values for home anticoagulation testing (HAT). Our study objectives were to assess this accuracy and compare the percentage time within therapeutic range (PTWTR) based on HAT data to that obtained with testing through an anticoagulation clinic service (ACS). Materials and methods: Forty-nine anticoagulated patients were enrolled in a year-long, prospective, crossover study comparing HAT to ACS testing. Patients performed HAT for 6 months and telephoned their INR values. Thereafter, devices were interrogated for the 30 most recent INR readings. Data accuracy was calculated for each patient as the percentage of correctly telephoned INR values divided by the total number of INR values common to both the device and the telephone logs. The device-based PTWTR was compared to the PTWTR based on ACS data. Results: Of the 49 enrolled patients, 32 completed the study protocol. The mean accuracy of reporting was 94.0 ± 13.0% (range, 48-100%); the median accuracy was 100%. Three patients had marked low accuracy (48, 60, 62%). No significant difference was seen between the PTWTR based on device data versus that obtained though the ACS (59.8 ± 15.7% versus 59.5 ± 19.4%, P = 0.48). Conclusions: The overall accuracy of patient-reported INR values is high and the PTWTR found with HAT is comparable to that obtained with clinic testing. However, the potential for noncompliance in a small number of patients raises the question of whether periodic confirmation of patient-reported INR values should be considered. © 2007 Elsevier Inc. All rights reserved.