Based on the impairment of erythropoietin production after allogeneic hematopoietic cell transplantation (HCT), we previously reported in a phase-2 trial that recombinant human erythropoietin (rhEPO) therapy was very efficient when started one month after transplantation. We also demonstrated that anemia after nonmyeloabalative (NM) HCT was less sensitive to rhEPO therapy than after conventional allogeneic HCT. This prompted us to confirm these findings in a prospective randomised trial. One hundred and thirty-one patients were randomised (1:1) between no treatment (arm 1) or erythropoietin (Neorecormon) at the dose of 500 U/kg/week (arm 2). Once the target Hb (13g/dL) has been attained, the dose of rhEPO was reduced by half, while it was withheld when Hb was = 14g/dL. Cohort A included 42 patients on day 28 after myeloablative HCT, cohort B 39 patients on day 28 after NMHCT, and cohort C 50 patients on day 0 of NMHCT. Primary endpoints included proportion of complete correctors (i.e. patients reaching Hb = 13g/dL) and median time to achieve Hb correction in each arm. The proportion of complete correctors before day 126 posttransplant was 0% in group 1A vs 52.4% in group 2A, 0% in group 1B vs 69.5% in group 2B and 19.1% in group 1C vs 70.2% in group 2C. Median time to achieve Hb = 13g/dL was not reached in group 1B vs 49 days in group 2B; 363 and 59 days in groups 1A and 1B respectively and 363 and 87 days in groups 3A and 3B respectively (figure 1). Hb evolution in each group is shown in figure 2. Seventyone patients (47/62 in control groups and 24/57 in treated groups, p=0.0003) required red blood cell transfusions. The difference was most pronounced in cohort B. There was no difference in rates of thrombo-embolic events or other complications between the two arms. In conclusion, this is the first trial to demonstrate that EPO therapy hastens erythroid recovery and decreases transfusion requirements when started one month after allogeneic HCT.