Patients with haematological malignancies undergoing allogeneic BMT were randomised to treatment with recombinant human erythropoietin (rHuEPO) (n = 25) or placebo (n = 25). rHuEPO was given at 200 U/kg daily for 4 weeks and 200 U/kg twice weekly for a further 4 weeks. The groups were similar regarding several prognostic factors. There were no differences between the two groups regarding time to engraftment, fever, hospitalisation, GVHD, infections, haemorrhages, transplant-related mortality, relapse and survival. However, more patients in the control group had a raised serum creatinine (43% vs 14%; p = 0.04). Red blood cell (RBC) transfusion requirements for the first 2 months after BMT were significantly lower in the rHuEPO group compared with the control group (5 units vs 10; p = 0.04). Time to unsupported Hb > 70 g/l was less in patients treated with rHuEPO (14 days vs 24; p = 0.03). No effect was seen on platelet engraftment or the number of transfused platelet units. Two patients in the control group compared with none in the rHuEPO group became refractory to platelet transfusions. According to the protocol the study drug was reduced (Hb > 100) or discontinued (Hb > 120) for a mean of 3.6 weeks among 11 rHuEPO patients compared with 1.9 weeks among 7 controls (p = 0.02). Seven of the treated patients compared with none of the controls reached Hb > 120 during the study period (p = 0.004). Among the rHuEPO treated patients, EPO-levels were significantly higher than in the controls.(ABSTRACT TRUNCATED AT 250 WORDS)