OBJECTIVE To compare the cost of recombinant human erythropoietin (rHuEPO) with that of blood transfusion in the treatment of chemotherapy-induced anemia from a healthcare system perspective. METHODS A decision analytic model. Baseline estimates were obtained from a review of clinical trials data and economic evaluation studies. METHODS Secondary data analyses of patients with advanced malignancies, excluding hematologic malignancies and metastasized solid tumors. METHODS Patients received either leukocyte-depleted packed red blood cells (PRBCs) or rHuEPO 150 units/kg s.c. three times per week for 6 months (24 wk). After 6 weeks, if rHuEPO recipients did not display a response, they received rHuEPO 300 units/kg s.c. three times weekly for the duration of therapy. If rHuEPO recipients still exhibited no response, they were given blood transfusions. RESULTS For a treatment period of 24 weeks, approximately 64% of rHuEPO recipients responded at an average expected cost of $12971 per patient. One hundred percent of transfusion recipients responded at a cost of $481; this resulted in a cost savings of $8490. Variation of response rates for rHuEPO or PRBCs did not appreciably lower costs. Lower rHuEPO dosages and higher numbers of transfused units of PRBCs yielded approximately equivalent costs; however, these strategies may not be clinically prudent. CONCLUSIONS From a healthcare system cost and outcome perspective, blood transfusion is the preferred strategy for chemotherapy-induced anemia. However, rHuEPO may be considered an effective blood-sparing alternative for patients with non-stem cell disorders. Future cost-effectiveness analyses are needed to assess more completely both the clinical and quality-of-life benefits rHuEPO may contribute to individual patients' lives and to society overall.