Twelve thromboplastins were tested against a secondary reference thromboplastin (human brain CRM BCR No. 147) or a tertiary house standard (human brain thromboplastin) with plasmas from normal healthy individuals and patients on oral anticoagulant therapy. The relationship between the prothrombin ratios of the thromboplastins tested versus the reference reagent was either a straight or curved line. The International Sensitivity Index (ISI) was estimated for each of the test thromboplastins and these ranged from 0.98 to 2.24. Some ISIs stated by manufacturers were different from our results. Thromboplastins with a high ISI showed a loss of sensitivity in assessing the level of anticoagulation at the upper end of the therapeutic range. In addition, the 95% Confidence Interval (CI) of the ISI estimations were widest for thromboplastins with the highest ISIs. Conversion of the prothrombin ratio to International Normalized Ratio (INR) was most accurate with the Australasian Reference Thromboplastin (ART) and least accurate with reagents having an ISI of 2.00 and over. Thromboplastins with an ISI between 1.10 and 1.50 may be adequate for the control of oral anticoagulant therapy, but were less accurate than a thromboplastin with an ISI approximating 1.00. Factors other than ISI should be considered in the choice of a thromboplastin, in particular a measurement of the accuracy of the ISI estimation such as the 95% confidence interval estimation used here.