The aim of this study was to assess whether the synthetic factor Xa inhibitor fondaparinux reduced the risk of venous thromboembolism more efficiently than the low molecular weight heparin dalteparin in patients undergoing major abdominal surgery. In a double-blind double-dummy randomized study, patients scheduled for major abdominal surgery under general anaesthesia received once-daily subcutaneous injections of fondaparinux 2.5 mg or dalteparin 5000 units for 5-9 days. Fondaparinux was started 6 h after surgery. The first two doses of dalteparin, 2500 units each, were given 2 h before surgery and 12 h after the preoperative administration. The primary outcome measure was a composite of deep vein thrombosis detected by bilateral venography and symptomatic, confirmed deep vein thrombosis or pulmonary embolism up until day 10. The main safety outcome measure was major bleeding during treatment. Among 2048 patients evaluable for efficacy, the rate of venous thromboembolism was 4.6 per cent (47 of 1027) with fondaparinux compared with 6.1 per cent (62 of 1021) with dalteparin, a relative risk reduction of 24.6 (95 per cent confidence interval -9.0 to 47.9) per cent (P = 0.144), which met the predetermined criterion for non-inferiority of fondaparinux. Major bleeding was observed in 49 (3.4 per cent) of 1433 patients given fondaparinux and 34 (2.4 per cent) of 1425 given dalteparin (P = 0.122). Postoperative fondaparinux was at least as effective as perioperative dalteparin in patients undergoing high-risk abdominal surgery.