The effects of therapy with recombinant human erythropoietin (Epoetin alfa) on erythropoiesis, preoperative autologous blood donation, and risk of exposure to allogeneic blood were evaluated in 204 patients scheduled to undergo elective orthopedic surgery. Study protocol required patients to have a baseline hematocrit < or = 39% and surgery scheduled 25-35 days in advance. Patients were randomized to two equal groups and were seen at study centers every 3-4 days within the 21-day trial period. At each visit, phlebotomy(< or = 450 mL) was performed if the hematocrit was > or = 33%, and Epoetin alfa (600 U/kg) or placebo was administered intravenously. A total of 173 patients were assessable; 31% of placebo recipients and 20% of Epoetin alfa recipients required allogeneic transfusion (p = 0.09). Logistic regression modeling showed that the risk of allogeneic transfusion was reduced by Epoetin alfa (p = 0.025). When patients receiving > 6 units of blood (necessitating allogeneic units) were excluded from analysis, 29% of placebo recipients and 14% of Epoetin alfa recipients were exposed to allogeneic blood (p = 0.015). Epoetin alfa recipients predonated more autologous units than did placebo recipients (4.5 vs 3.0 units, respectively; p < 0.001), and their production of red blood cells increased significantly more over baseline production values (668 vs 353 mL, respectively; p < 0.05). These results demonstrate that administration of Epoetin alfa stimulates erythropoiesis, allows predonation of more units of autologous blood, and reduces the risk of exposure to allogeneic blood. Optimal dosing regimens and surgical patients most likely to benefit fro Epoetin alfa therapy must be established.