To compare maintenance epoetin alfa administered once every 3 weeks with continued weekly epoetin alfa for patients with cancer-associated anemia. Eligible patients were randomly assigned at enrollment to receive three weekly doses of epoetin alfa 40,000 U subcutaneously (SC), followed by either standard weekly epoetin alfa (40K arm) or 120,000 U of epoetin alfa (120K arm) SC every 3 weeks for 18 additional weeks. Three hundred sixty-five patients were enrolled. One hundred eighty-three patients were assigned to the 40K arm, and 182 were assigned to the 120K arm. There was no difference in the proportion of patients requiring transfusions during the study (23% in 40K arm and 18% in 120K arm, P = .22) or specifically during the maintenance phase (13% in 40K arm v 15% in 120K arm, P = .58). Patients randomly assigned to the 40K arm were more likely to have a > or = 2 or > or = 3 g/dL hemoglobin (Hb) increment, were more likely to have a drug dose held because of high Hb, and had higher mean end-of-study Hb levels. Toxicities, including thromboembolism, and overall survival were similar. Patients in the 40K arm had a higher global quality of life (QOL) at baseline for unclear reasons, whereas patients in the 120K arm had a greater global QOL improvement during the study, so end-of-study QOL was equivalent. After three weekly doses of epoetin alfa 40,000 U, a dose of 120,000 U can be administered safely once every 3 weeks without increasing transfusion needs or sacrificing QOL. The Hb increment is somewhat greater with continued weekly epoetin alfa. Lack of blinding as a result of different treatment schedules may have confounded results.