The contribution of the thromboplastin international sensitivity index (ISI) to the interlaboratory coefficient of variation (CV) of the international normalised ratio (INR) with individual reagents was assessed. In theory the precision of the INR should increase with lower ISI values. An empirical relation has been established between the ISI, the INR, and its CV for two rabbit thromboplastins used in sufficient numbers for analysis in the United Kingdom. This was based on the cumulative data from the United Kingdom National External Quality Assessment Scheme (NEQAS) surveys over two years beginning in 1986. The actual precision achieved in NEQAS for the two reagents depends on the ISI value of the thromboplastin and it agreed closely with the figure predicted by the empirical model. The findings show that the ISI value of a thromboplastin strongly influences the interlaboratory variability of the INR obtained with it. The CV of the INR approximates to the CV of the prothrombin ratio multiplied by the ISI. Manufacturers of thromboplastin should therefore be encouraged to produce reagents showing good precision of results reported as simple prothrombin ratios and a low ISI value to avoid impairment in precision when ratio results are transformed to INR.