Summary The purpose of this study was to determine the safety, efficacy and impact on quality of life of recombinant human erythropoietin (r-HuEPO) for cancer patients undergoing radiotherapy (RT). An open-labelled randomized design was used, with patients randomized to either treatment or control arms. Patients in the treatment arm received r-HuEPO given by subcutaneous injection at a dose of 200 units kg-' day-' plus oral iron supplements (ferrous sulphate 325 mg p.o. t.i.d.). Entry was restricted to patients with carcinoma of the lung, uterine cervix, prostate or breast who presented for RT with anaemia parameters reflective of 'the anaemia of chronic disease'. Radiotherapy policies (portals, doses, fraction size, etc.) were determined by the site and stage of disease. Complete blood counts (CBCs) were obtained weekly. The target level of haemoglobin was 15 g dl-1 for men and 14 g dl-1 for women. Quality of life (QOL) was assessed weekly by using an analogue scale to judge energy, activities of daily living and overall quality of life. Forty-eight patients were entered in the study, 24 in the treatment arm and 24 in the control arm. The prerandomization demographic characteristics and mean laboratory values were comparable in both arms. The mean haemoglobin at completion was 13.6 g dl-1 for r-HuEPO-treated patients compared with 11.0 g dl-1 for control subjects (P= 0.0012). Patients who received r-HuEPO demonstrated a mean weekly haemoglobin increase of 0.41 g dl-1 compared with a decrease in mean haemoglobin level in controls for 6 of the 7 weeks of the study (mean weekly decrease of 0.073 g dl-'). Target levels of haemoglobin were achieved by 41.6% of r-HuEPO-treated patients compared with none of the control subjects. The mean platelet count declined in both arms of the study with RT but the decline from pretreatment was less rapid in r-HuEPO-treated patients (11.2% decrease) compared with controls (26.3% decrease) and was statistically significant during weeks 4-6. Toxicity was minor with only mild irritation at the injection site. Mean quality of life end points were superior in the treatment arm but not statistically significant. r-HuEPO had a beneficial effect on weekly haemoglobin levels in patients undergoing RT with response rates similar to other studies. There was also a less rapid decline in weekly platelet counts in r-HuEPO-treated patients compared with control subjects. Further studies are needed to address the optimum dose and scheduling as well as the impact of r-HuEPO on clinical outcomes.