Venous thrombosis rates were compared in 200 patients undergoing total hip arthroplasty and randomized to receive either fixed mini-dose warfarin (1 mg daily) or adjusted-dose warfarin to maintain an international normalized prothrombin ratio (INR) of 2.0-4.0. Bilateral lower limb venography was performed between days 11 and 13 inclusive. Fixed mini-dose warfarin was associated with a significantly higher rate of total thrombosis (P less than 0.05). General anaesthesia was associated with a significantly higher rate of thrombosis than spinal anaesthesia (P less than 0.05). Adjusted-dose warfarin was associated with more bleeding complications than mini-dose warfarin although these were not attributable to excessive anticoagulation. A single death from pulmonary embolus occurred in the early postoperative period in a patient receiving adjusted-dose warfarin.