The effect of recombinant human erythropoietin (EPO) on hospitalization of patients with end-stage renal disease (ESRD) was evaluated in a controlled clinical trial. A cohort of 67 new hemodialysis patients prescribed EPO shortly after the clinical availability of EPO were the treatment group. The control group was a cohort of 67 new hemodialysis patients matched for clinical center, age, cardiovascular disease and transfusion history. These patients had not been prescribed EPO as they had started hemodialysis prior to the clinical availability of EPO. There were 21 pairs without hospitalization and 46 pairs with at least 1 member of the pair experiencing hospitalization. Among the latter group, the median follow-up was 174 and 184 days for the EPO and control patients respectively. For all hospitalizations, those treated with EPO were hospitalized 15.3 days per year compared to 23.2 days for the control patients. The difference (EPO-control) was -7.9 days (95% CI: -21.0; 7.8) for all cause hospitalization. For hospitalizations due to cardiac, infectious disease and gastrointestinal disease, the differences were 1.6, 1.8 and 1.2 days favouring EPO treated patients. For hospitalizations related to vascular access complications, the difference was 0.9 days favoring the control group. All other causes favoured EPO treated patients by 4 days. There had been 58 hospitalizations in the EPO group compared to 97 in the control group. The mean duration of hospitalization was 8.0 days for the EPO and 9.6 for the control group. The direction and magnitude of the change in all cause hospitalization represents an improvement in morbidity and an important decrease in health resource utilization.(ABSTRACT TRUNCATED AT 250 WORDS)