To assess the efficacy and the optimum dose of recombinant human erythropoietin (rhEpo) on the anemia of premature, 45 preterm infants with a gestational age of less than 35 weeks and birth weight of less 1 800 g were randomly assigned to treatment group 1 (n = 15, receiving subcutaneous rhEpo 150 U/kg · time), treatment group 2 (n =15, receiving 250 U/kg · time), three times a week for 6 weeks, and control group (n =15, no treatment was given). All preterm infants received supplements of vitamin E (20 IU) and iron (20 mg) each day. Our results showed that postnatal decline of hemoglobin (Hb) and hematocrit (Hct) were lessened in the treatment groups, particularly in the group 2 and the differences were very significant (P<0.0001 for all). Treated infants had significantly higher reticulocyte counts (Ret) (P<0.0001 for all), but there was no significant difference between the two treatment groups (P > 0.05). Serum iron dropped significantly in the treatment groups as compared with control group (P<0.01 for all), but no dose-dependent relationship was observed in treated infants (P>0.05). After treatment, serum levels of erythropoietin was higher in group 2 than those in group 1 and control group (P<0.0001,P<0.01 andP<0.05, respectively). There was no significant difference between group 1 and control group (P>0.05). No side effects related to rhEpo therapy were observed. Our study suggested that rhEpo therapy stimulates endogenous erythropoiesis and enhances Ret, Hct and level of Hb in a dose-dependent manner in premature infants. The therapy is more efficient when given in higher dosages.