AIMS: To assess the accuracy and precision of INR measurement by trained practice and district nursing staff using the Thrombolytic Assessment System (TAS) analyser. METHODS: Seventeen nurses from four practices were trained to measure INR using the TAS analyser on citrated capillary blood samples. Quality control (QC) consisted of: daily internal QC using normal and abnormal commercial plasmas; monthly local external QC scheme using fresh citrated venous blood; and registration of all analysers in the NEQAS (national external quality assessment scheme) main users scheme. RESULTS: Analysis of internal QC results demonstrated satisfactory interanalyser and intra-analyser precision with no evidence of analytical drift in any of the four practice analysers over an eight month period. Local and national external QC results confirmed the interanalyser precision but INR was underestimated by the TAS analysers compared with the CA 1000 using either Diagen rabbit brain thromboplastin or Innovin, and with other NEQAS users. CONCLUSIONS: The TAS analyser has many features to commend it for use by nonpathology staff to determine INR. Local internal and external QC and entry into the NEQAS main users group are possible because the TAS analyses citrated plasma or blood. The TAS analyser underestimates INR when the geometric mean normal prothrombin time (GMNPT) is determined by conventional methods. A local correction factor can be introduced by adjusting the normal PT to give INR results comparable with the local laboratory. This is particularly desirable when INRs are measured using both near-patient and laboratory analytical systems on different occasions.