MDS are a heterogeneous group of clonal disorders characterized by the development of one or more symptomatic cytopenias. We evaluated the role of erythropoietin (EPO) with or without granulocyte colony stimulating factor (G-CSF) for the treatment of anemic patients with MDS. Patients with refractory anemia (RA n=40), RA with ringed sideroblasts( RARS n=36), and RA with excess blasts (RAEB n= 29) were initially randomized to either supportive therapy (ST) or EPO, 150 u/kg/d. Patients on the ST arm could cross over to the EPO arm if after at least a 4 month period of observation they had a ≥ 50% increase in their transfusion requirement. In patients who did not respond to or did not maintain their response to EPO( 150 u/kg) , G-CSF, 1mcg/kg/d was administered in addition to the EPO. Patients who did not respond after 4 months received an increased dose of EPO at 300u/kg/d + G-CSF. Bone marrow studies and iron stores were evaluated before each change in treatment. Response criteria were CR (complete response), GR (good erythroid response), which required a sustained increase of the pre-transfused Hgb value of > 2 g/dl or loss of the transfusion requirement or PR (partial response), defined as a decrease in the transfusion requirement by ≥ 50% or an increase in the Hgb by > 1 g/dl above the pretreatment level. A GR or PR must be sustained for at least 4 months. Transformation to acute leukemia was defined as ≥ 30% blasts in the bone marrow. 118 patients were enrolled , 109( 53 EPO, 56 ST) were eligible and 105 were evaluable for response. The median age was 73 (range 37–88 years). The response rate (CR+GR+PR) in the EPO arm was 35% vs. 9% in the ST arm (p=. 002). Of the 23 patients who crossed over from the ST arm, 30% responded to EPO (GR+PR). Six of 27 patients (22%) who received EPO 150u/kg + G-CSF responded (CR+GR+PR). Ten patients received EPO 300 u/kg + G-CSF and 6(60%) responded (CR+GR+PR). Transformation to acute leukemia occurred in 3.6% and 0% in the supportive therapy and EPO arm respectively( P=.50). A complete response (CR) was documented in 2 patients, 1 each on EPO and EPO+G-CSF. There was a survival difference for patients who had an erythroid response: median survival 53 months for responders vs 26 months for non-responders (p=.009). Neither EPO nor the addition of G-CSF was associated with an increase rate of transformation to acute leukemia. Toxicities were comparable in all the treatment arms. The pretreatment serum EPO level correlated with the response to treatment (P=.004): median EPO of 48 vs. 140 mu/ml for responders versus non-responders. Responses were observed in all subtypes. Responses were greater in RA>RARS>RAEB. In conclusion, EPO administration was associated with a significant increase in Hgb and a decrease in the transfusion requirement in anemic patients with MDS. Low doses of G-CSF + EPO were effective in EPO nonresponsive or refractory patients. In MDS patients EPO + low dose G-CSF was not associated with progression or transformation to acute leukemia.