Anemia (characterized by a hematocrit of 30 percent or lower) persists in 40 to 60 percent of patients treated for end-stage renal disease with maintenance hemodialysis, despite concomitant erythropoietin (epoetin) therapy. We tested the hypothesis that inadequate dialysis is a key reason for the insufficient response to erythropoietin in patients with end-stage renal disease who are receiving hemodialysis. We prospectively studied 135 randomly selected patients undergoing hemodialysis who had been receiving intravenous erythropoietin for at least four months. The adequacy of dialysis was assessed by measuring the percent reduction in the blood urea nitrogen concentration and the serum albumin concentration. The hematocrit was measured weekly for four weeks, transferrin saturation was measured, and coexisting illnesses were documented. To determine the effect of an increased level of dialysis on the hematocrit, the thrice-weekly schedule of dialysis was increased to raise the mean urea-reduction value from 60.7 to 72 percent for six weeks in 20 consecutive patients whose base-line urea-reduction value was less than 65 percent. The change in the hematocrit in these patients was compared with that observed in the next 20 patients who had an equivalent base-line urea-reduction value but whose level of dialysis was not altered. The mean hematocrit of the entire group was 29.2 +/- 4 percent, and the mean thrice-weekly dose of erythropoietin was 59 +/- 29 U per kilogram of body weight. The mean serum albumin concentration was 3.8 +/- 0.4 g per deciliter, the mean urea-reduction value was 62 +/- 4.8 percent, and the mean transferrin saturation was 20 +/- 9 percent. Multiple regression analysis revealed direct correlations between the hematocrit and the serum albumin concentration (P = 0.009) and between the hematocrit and the urea-reduction value (P = 0.012) after adjustment for other factors. A logistic-regression analysis indicated that an 11 percent increase in the urea-reduction value doubled the odds that a patient would have a hematocrit above 30 percent. After six weeks of increased intensity of dialysis in 20 patients with base-line urea-reduction values of less than 65 percent, the mean (+/- SE) hematocrit rose from 28.4 +/- 0.78 percent to 32.3 +/- 0.71 percent (P = 0.002); there was no significant change in a control group of 20 patients with equivalent base-line urea-reduction values in whom the dialysis level was not altered (28.2 +/- 0.84 percent to 26.3 +/- 0.85 percent; P = 0.175). In patients with end-stage renal disease, inadequate hemodialysis is associated with a suboptimal response to erythropoietin therapy. Increasing the intensity of dialysis in patients with anemia who are receiving inadequate dialysis results in a significant increase in the hematocrit.