The limitation to exercise capacity in hemodialysis patients has been attributed to anemia. We report the effects of normalization of hematocrit levels by using r-hu-recombinant erythropoietin and exercise training on exercise capacity and self-reported physical functioning in hemodialysis patients. Sixty-five patients were randomized into 1 of 4 groups: usual hematocrit (30%-33%) with no exercise training (UH); usual hematocrit (30%-33%) plus exercise training (UHX); normalized hematocrit (40%-42%) with no exercise training (NH); and normalized hematocrit (40%-42%) plus exercise training (NHX). Treadmill exercise testing was conducted at baseline and at 5 months after the initiation of the interventions. Analysis was performed on the data collapsed for 48 patients who met the criteria for hematocrit and exercise adherence and completed both baseline and post intervention (5.6 +/- 1.6 months) testing. Significant effects of exercise were found in peak oxygen uptake measurements (P = 0.03) and in self-reported physical functioning as measured by the Short Form-36 questionnaire (P = 0.01). There was a significant effect of hematocrit on the General Health scale on the SF-36 (P = 0.03). The changes in peak oxygen uptake with exercise training were small and levels remained lower than age-predicted values at the end of the study. These results indicate that there are other physiologic limitations to exercise capacity that are not overcome by exercise training or normalization of hematocrit. The effects of exercise training on self-reported physical functioning may be of clinical importance because these scores have been shown to be highly predictive of outcomes such as hospitalizations and mortality in hemodialysis patients.