Thromboplastins vary in their sensitivity to the haemostatic defect induced by oral anticoagulants. To provide a means of standardising prothrombin time tests, the World Health Organization adopted in 1977 a scheme for calibrating thromboplastins in terms of an International Reference Preparation. Unfortunately, the model on which this scheme was based does not always hold. A revised calibration model has therefore been developed and this has been tested in a recent collaborative study. The revised model, which retains fundamentally the same principle for standardising prothrombin time tests, has proved suitable for calibrating thromboplastins of different species and types and, moreover, has certain statistical advantages over its predecessor. In September 1982, the WHO Expert Committee on Biological Standardization adopted the revised model. This paper explains the nature and rationale of this change and considers its practical implications.