Background. Monitoring of oral anticoagulant therapy (OAT) is usually accomplished by measuring prothrombin time and the international normalized ratio (INR). However, thromboplastins have different responsiveness and sensitivity to vitamin K-dependent coagulation factors depletion. Several studies have shown INR variation when low sensitive thromboplastins are used. This study compared INR variability between two laboratories using highly sensitive thromboplastins. Methods. A total of 237 plasmas were tested, half of them from patients under OAT. Samples were tested simultaneously in two laboratories: in laboratory A, a Behring Coagulation Timer instrument and a human recombinant thromboplastin (Innovin, Dade Behring) (ISI 1.01) were used. In laboratory B, a Thrombolyzer Compact (Behnk Elektronik) and a rabbit brain thromboplastin (Simplastin Excel S, Organon Teknika) with an ISI of 1.30 were used. Statistical analysis was carried out according to the method of Bland and Altman. Results. Even though high correlation coefficients were obtained when comparing both laboratories, Bland-Altman analysis showed a variation of INR between laboratories ranging from -0.77 to +1.07. After logarithmic transformation of data, these values yielded a variation of the INR either 25% below or 44% above. Conclusions. These results are clearly inadequate for clinical use because such a variation would most probably induce the clinician to make a change in warfarin dose. Standardization of instruments, reagents, and controls is warranted to decrease this variation.