OBJECTIVE--To evaluate the efficacy and safety of two subcutaneous prophylactic regimens for postoperative deep vein thrombosis after total hip replacement. DESIGN--Prospective open randomised multicentre trial. SETTING--28 European departments of orthopaedic surgery. INTERVENTION--All patients had bilateral phlebography 10 days after surgery. 31 patients receiving low molecular weight heparin and 29 receiving unfractionated heparin were excluded from the efficacy analysis for various reasons. PATIENTS--349 patients undergoing total hip replacement between September 1988 and May 1989. 174 patients received subcutaneously a low molecular weight heparin (Fraxiparine) with anti-factor Xa activity of 41 IU/kg/day for three days, then 62 IU/kg/day from day 4 to day 10. 175 patients received subcutaneous unfractionated heparin at intervals of eight hours; doses were adjusted to maintain the activated thromboplastin time at two to five seconds above control values. MAIN OUTCOME MEASURE--Total incidence of deep vein thrombosis and incidence of proximal deep vein thrombosis on bilateral phlebography. RESULTS--The total incidence of deep vein thrombosis was 16% in patients receiving unfractionated heparin and 12.6% in patients receiving low molecular weight heparin (p = 0.45), and the incidence of thrombosis of the proximal veins was 13.1% and 2.9% respectively (p less than 0.001). Four patients receiving unfractionated heparin and one receiving low molecular weight heparin developed pulmonary embolism. The incidence of bleeding complications was low and comparable in the two groups. CONCLUSION--Low molecular weight heparin is at least as effective as unfractionated heparin in preventing deep vein thrombosis and is more effective at preventing thrombosis of the proximal veins in patients undergoing hip replacement. Low molecular weight heparin is not more likely to cause bleeding complications and is simpler to give than unfractionated heparin.