BACKGROUND: The optimum regimen of epoetin alfa for prevention of allogeneic blood transfusion is unknown. OBJECTIVE: To determine whether a modified regimen of epoetin alfa reduces allogeneic blood transfusion in patients undergoing hip arthroplasty. DESIGN: Randomized, double-blind, multicenter trial comparing two modified dose regimens of epoetin alfa with placebo. SETTING: 13 teaching hospitals and 4 community hospitals in Canada. PATIENTS: 201 patients undergoing primary hip arthroplasty who had a hemoglobin concentration of 98 to 137 g/L and did not predonate blood. INTERVENTION: Patients were assigned in a 3:5:5 ratio to receive four weekly doses of epoetin alfa, 40 000 U (high-dose; n = 44) or 20 000 U (low-dose; n = 79), or placebo (n = 78), starting 4 weeks before surgery. All patients received oral iron supplementation, 450 mg/d, for 42 or more days before surgery. MEASUREMENTS: The primary end point was allogeneic transfusion. Secondary end points were thromboembolic events and change in reticulocyte count and hemoglobin concentration. RESULTS: Both modified epoetin alfa regimens significantly reduced the need for allogeneic transfusion: Five (11.4%) patients in the high-dose group (P = 0.001) and 18 (22. 8%) patients in the low-dose group (P = 0.003) had transfusion, compared with 35 (44.9%) patients in the placebo group. The hematologic response was substantial in patients who received epoetin alfa. In the high-dose group, low-dose group, and placebo group, the preoperative increase in reticulocyte count was 58.8, 37. 0 and 1.8 x 10(9) cells/L (P < 0.001), respectively, and the increase in hemoglobin concentration was 19.5, 17.2, and 1.2 g/L (P < 0.001). The incidence of thromboembolic events did not differ among groups. CONCLUSIONS: Both modified epoetin alfa regimens were effective compared with placebo in reducing allogeneic transfusion in patients undergoing hip arthroplasty. Patients who received high-dose epoetin alfa had the lowest transfusion rate.