Bleeding can occur unexpectedly during antithrombotic therapy. Impaired haemostasis is commonly measured by the bleeding time. We measured it by 3 methods in controls and in anticoagulated animals and related it to their antithrombotic status. In 42 control rats template, tail-tip transection and needle occlusion bleeding times correlated poorly (r=0.05-0.34). The template method had the best range (mean 126.97±SEM sees) and consistency. In 10 control animals it correlated mildly (r=0.55) with venous thrombus in the same animal. Thrombus was measured by its weight deposited on platinum wires (2 cm long, 0.4 mm diameter) set in vein and in artery for 1 h. In respective groups of 10 rats, a decrease of mean thrombogenesis was obtained using aspirin, heparin and low molecular weight heparin in 2 dosages and hirudin in 1 dosage. The drugs reduced mean venous thrombus by 13-86 per cent of the mean control thrombus, and prolonged the mean template bleeding time by 29-199 per cent. The ranking of the drugs according to their increase of template bleeding time was virtually the same as the ranking given by their reduction of thrombus weight (Spearman rank coefficient 0.81, sig 0.007). The transection test produced a similar ranking and similar correlation with thrombus (0.71, sig 0.049). Low molecular weight heparin induced the greatest thrombus reduction (39 per cent) for least prolongation of bleeding time (24 per cent). Arterial thrombus was more variable. The bleeding times and thrombus weight were measured in each animal of 2 groups given aspirin, the template method correlating mildly with venous thrombus reduction (r=0.23, 0.58 respectively), the transection method with arterial (0.74, 0.45) and the occlusion test poorly with either (0.13, 0.22). Bleeding time lengthens with increasing antithrombotic effect of drugs, but not in direct proportion, nor similarly with each drug.