Objective - To assess the cost benefits of low dose subcutaneous recombinant human erythropoietin in correcting the anaemia of end stage renal disease. Design - Three year retrospective study. Setting - Subregional nephrology service serving a mixed urban and rural population of 800,000. Subjects - 60 patients with symptoms of anaemic end stage renal disease treated with erythropoietin (43 receiving haemodialysis; 11 receiving continuous ambulatory peritoneal dialysis; two with predialysis end stage renal disease; four with failing renal transplants). Main outcome measures - Costs and savings of achieving and maintaining an haemoglobin concentration of 85-105 g/l with erythropoietin. Results - All patients treated with erythropoietin achieved the target haemoglobin concentration at median induction doses of 97 (95% confidence interval 95 to 108) units/kg/week, and this was maintained with 79 (75 to 95) units/kg/week at an average annual cost per patient of £ 2260. Admissions related to anaemia were virtually eliminated (246 v 1 inpatient days for 12 months before and after starting erythropoietin). 54 patients required no blood transfusions after starting erythropoietin, and the total requirements fell from 230 to 21 units in the 12 months before and after starting erythropoietin. Iron stores were maintained with oral or intravenous iron. All patients reported increased wellbeing, appetite, and exercise capacity. Hypertension developed or worsened in 30 patients, resulting in hospital admissions in five patients, one of whom had seizures. Conclusion - Low dose subcutaneous erythropoietin restores haemoglobin concentrations sufficiently to abolish blood transfusion requirements and reduce morbidity. The net cost of erythropoietin prescribed in this way (£ 2260/patient/year) was largely offset by savings in costs of hospital admissions. The true annual cost to the NHS was around £ 1200 per patient.