The dosage of the anticoagulant warfarin sodium is based upon the prolongation of the prothrombin time into an optimal therapeutic range. We have developed a new assay for the native prothrombin antigen that measures the fully gamma-carboxylated prothrombin using a radioimmunoassay. Based on preliminary data that indicated that the native prothrombin antigen predicted both bleeding and thrombotic complications more accurately than the prothrombin time in patients anticoagulated with warfarin sodium, we have performed a randomized prospective trial comparing the complication rate in warfarin-treated patients monitored with the native prothrombin antigen or the prothrombin time. Patients with indications for anticoagulation were randomized to be monitored by the native prothrombin antigen (therapeutic range, 12 to 24 micrograms/mL) or the prothrombin time index (therapeutic range, 1.5 to 2.0). Of the prothrombin time group (N = 80), seven (8.8%) had bleeding or thrombotic complications, with a complication rate of 9.5%/patient-year. In the native prothrombin antigen group (N = 76), one subject (1.3%) had a bleeding complication. The complication rate per patient-year was 1.5%. These results indicate an 85% reduction in the complication rate of the native prothrombin antigen group compared with the complication rate of the prothrombin time group. This difference is statistically significant by the Fisher exact test (P = .037) and by Kaplan Meier survival analysis (P = .040). This study suggests that the use of the native prothrombin antigen assay has the potential to decrease the complications associated with anticoagulation therapy with warfarin sodium.