Twenty patients with myelodysplastic syndromes (MDS) entered a randomized, placebo-controlled, double-blind trial designed to evaluate the efficacy and toxicity of high doses of recombinant human erythropoietin (rhEPO). Patients completing the trial were eligible to receive rhEPO as part of an open-label study. Eighteen patients were transfusion dependent; 10 had refractory anemia (RA), and 10 had refractory anemia with ringed sideroblasts (RARS). A response to rhEPO was defined as an increase in hematocrit of 4 percentage points or more over baseline, or the elimination of all transfusions with the hematocrit stable at the baseline level. In the double-blind trial, 1 patient (12.5%) receiving rhEPO responded, as compared with no responses in the placebo group. Overall, responses occurred in 4 of 17 patients (24%) receiving rhEPO at a dose of 1,200 to 1,600 U/kg intravenously (IV) twice weekly. Changes in granulocyte or platelet counts were not observed. Despite the administration of high doses of rhEPO, toxicity attributable to rhEPO was not observed in either the double-blind or open-label study. Response to rhEPO was not significantly related to age, gender, type of MDS, time since diagnosis, time since initiation of transfusion therapy, or baseline serum EPO. These studies indicate that rhEPO can be administered safely in very high doses to patients with MDS and that 24% of these patients will respond with increased erythropoiesis.