An international collaborative study involving 20 laboratories was undertaken to evaluate the status of standardization of a commercial recombinant thromboplastin. The results show that the reagent fulfills the WHO requirements for thromboplastin calibration against two International Reference Preparations (IRP) from different species, because there was a linear relationship of log-transformed prothrombin times as measured by the recombinant reagent and the two IRPs in normal individuals and patients stabilized on oral anticoagulants. The regression lines drawn through the patient data points passed through the normal data points in the majority of instances. The average ISI values for three different batches were close to unity. On the average, the ISI value calculated against BCT/253 (human, plain) was slightly smaller than that against RBT/79 (rabbit, plain). The between laboratory precision of calibration (CV = 5-6%) did not change whether BCT/253 or RBT/79 was used, whereas it was substantially improved when the calibrations of batches 2 and 3 were performed against batch 1, arbitrarily taken as standard (CV < or = 1.7%). Statistical analysis to test for differences between slopes revealed no significant between-batch differences in 16 of 20 laboratories. However slight differences for the third batch (2.8% less than the average ISI value of batches 1 and 2) leave room for further improvement in the standardization of the reagent.