Current performance in the prothrombin time (PT) of the two main United Kingdom coagulometer/thromboplastin systems was assessed in a field survey. Twenty abnormal samples covering a wide spectrum of International Normalised Ratio (INR) were distributed to users of the KC4/KC10 and Coag-a-Mate instruments. Coagulometer results were compared with those of the manual method. A substantial minority with each system showed good agreement with the manual reference values. There was, however, a considerable variation between instruments, meaningful in clinical terms, evidenced by varying regression slopes and local system International Sensitivity Indices (ISI). For intense anticoagulation (3.0 to 4.5 INR) a larger dose of warfarin is needed with the Coag-a-Mate than with the KC instruments. With a manual INR of 4.0 the KC instruments tended to give longer PT (mean INR + 0.3); the Coag-a-Mate PT was generally shorter (mean INR -0.1). With both systems the mean normal PT were shorter than the manual but the degree of shortening did not parallel that of the abnormal samples. This effect undermines the use of a simple prothrombin ratio and of an INR value derived from it, based on a manual ISI. The use of a system related ISI cannot, however, be recommended until local instrument variables are controlled.