Twenty-four premature infants, < 32 weeks gestational age, were randomly assigned in a double-blind, placebo-controlled trial to 6 weeks of treatment with either recombinant human erythropoietin (rHuEpo) 150 U/kg three times per week given sc (n = 12) or placebo (n = 12). The infants were fed a diet rich in protein (3.2 g/kg/day) and energy (130 kcal/kg/day) based on their own mother's milk fortified with bovine protein together with moderate iron supplementation (4 mg/kg/day). During the treatment (rHuEpo versus placebo) significant differences in mean (+/- SD) reticulocyte count (4.8 +/- 1.2 versus 2.7 +/- 1.4%; p < 0.01), mean packed red cell volume (PCV) (0.38 +/- 0.03 versus 0.34 +/- 0.04, p < 0.05) and mean haemoglobin concentration (12.6 +/- 1.1 versus 11.5 +/- 1.2 g/100 ml; p < 0.05) were found. Within the rHuEpo group, PCV and haemoglobin concentration remained unaltered from entry to 1 week after cessation of treatment whereas a significant decline was observed in the placebo group. No indications of iron deficiency were seen. We conclude that moderate doses of rHuEpo given to infants fed a diet rich in protein and energy are effective in ameliorating anaemia of prematurity. High iron supplementation does not seem to be essential for a significant erythropoietic response. No adverse effect attributable to rHuEpo was observed.