In part A of this study, patients were randomised to cohorts receiving darbepoetin alfa at doses of 0.5 to 8.0 m.c.g x kg(-1) x wk(-1) or to a control group receiving epoetin alfa at an initial dose of 150 U x kg(-1) three times weekly. In part B, the cohorts were darbepoetin alfa 3.0 to 9.0 m.c.g x kg(-1) every 2 weeks or epoetin alfa, initial dose 40 000 U x wk(-1). Safety was assessed by adverse events, changes in blood pressure, and formation of antibodies to darbepoetin alfa. Efficacy was assessed by several haematologic endpoints, including change in haemoglobin from baseline. The adverse event profile of darbepoetin alfa was similar to that of epoetin alfa. No relationship between the rapidity of haemoglobin response and any adverse event was observed. No antibodies to darbepoetin alfa were detected. Higher doses of darbepoetin alfa increased the proportion of patients with a haemoglobin response and decreased the median time to response. The overall dose of darbepoetin alfa required to produce a mean increase in haemoglobin does not increase when the dosing interval is increased from 1 to 2 weeks. Therapy with darbepoetin alfa is safe and effective in producing a dose-related increase in haemoglobin levels in patients with cancer receiving chemotherapy.