AIMS: To examine the reliability of international normalised ratio (INR) determination on samples stored as whole blood for up to two days at room temperature. METHODS: The INR of 40 patients receiving oral anticoagulants was determined on fresh blood and on samples stored for 24 and 48 hours, using five locally calibrated prothrombin time systems. These incorporated Manchester reagent, Recombiplastin, IL PT Fibrinogen HS Plus, Manchester combined capillary prothrombin time reagent, and a freeze dried in-house reference rabbit brain thromboplastin, RBT 1010. In addition, factors II, V, VII, and X were determined on samples obtained from 18 of these patients before and after incubation at room temperature. RESULTS: The INR of the samples changed by differing amounts during storage, depending on which system was employed. Although the mean change after 24 hours storage was relatively small, there were individual samples that changed by > 0.5 INR with all systems. These changes would lead to adjustment in dosage of certain patients. After 48 hours these effects were greater with all systems except that employing Recombiplastin. There were only small reduction in the measured factors by 48 hours. CONCLUSIONS: After storage of samples for only 24 hours, some patients' INR changed sufficiently to affect dosage. In view of these observations, the practice of storing whole blood samples for INR determination cannot be recommended.