PURPOSE: To assess physician use of erythropoietin in cancer patients before publication of the American Society of Clinical Oncology/American Society of Hematology guidelines. METHODS: Questionnaires about erythropoietin use in practice and 12 hypothetical clinical scenarios involving patients with cancer were mailed to 2000 oncologists/hematologists in the United States and 19 other countries. Response rates were 30% in the United States and 25% internationally. Data on erythropoietin use for ovarian cancer were obtained from one clinical trial. Multivariate regression models assessed predictors of erythropoietin prescription. RESULTS: Most physicians selected a hemoglobin level < or =10 g/dL as an upper threshold for erythropoietin use (36% to 51% of U.S. physicians and 21% to 32% of foreign physicians). Frequent erythropoietin use (defined as use in at least 10% of cancer patients) was higher in the United States than elsewhere (adjusted odds ratio [OR] = 5.8; 95% confidence interval [CI]: 2.5 to 13.4). Among U.S. physicians, those who said they used erythropoietin frequently were more likely to be in fee-for-service than managed care settings (OR = 2.2; 95% CI: 1.3 to 3.7). Those who reported never using erythropoietin practiced in countries that had lower annual per capita health care expenditures, lower proportions of privately funded health care, and a national health service (P <0.05 for all comparisons). Of 235 ovarian cancer patients who received topotecan, 38% (45/118) of U.S. patients and 2% (2/117) of European patients who developed grade 1 anemia (hemoglobin level between 10 and 12 g/dL) were treated with erythropoietin (P <0.01). CONCLUSION: Financial considerations and a hemoglobin level <10 g/dL appear to influence erythropoietin use in the United States, whereas financial considerations alone determine erythropoietin use abroad.