Early detection of iron sufficiency at the level of the erythropoietic cell is necessary to optimize management of uremic anemia with recombinant human erythropoietin (rHuEPO). "Absolute" and "functional" iron deficiency are the most important factors causing resistance to administered rHuEPO. Transferrin saturation and serum ferritin measurements have been noted to be insensitive and inaccurate measures to detect functional iron deficiency. Recently, the reticulocyte hemoglobin content (CHr) has been shown to be a sensitive and specific indicator of functional iron deficiency in nondialysis patients treated with rHuEPO. The purpose of this study is to compare CHr with currently used indices of iron sufficiency in rHuEPO-treated hemodialysis (HD) patients. In study 1, 364 stable HD patients were studied at two outpatient dialysis centers. CHr was normally distributed, with a mean value of 28.3 pg, and was consistent over two consecutive monthly samples in each center. CHr was weakly but consistently correlated with transferrin saturation and serum ferritin. CHr and reticulocyte number were inversely correlated with red blood cell (RBC) number, suggesting that the erythropoietic stimulus of routinely administered rHuEPO may have resulted in functional iron deficiency. Month-to-month changes in CHr correlated weakly with changes in serum iron and percent transferrin saturation, but not at all with changes in serum ferritin. When we analyzed those patients with baseline CHr less than 26 pg, a level strongly suggestive of functional iron deficiency, these correlations strengthened, and in addition, month-to-month changes in CHr correlated strongly and directly with concomitant changes in RBC count, hemoglobin, and hematocrit, suggesting that rising CHr was indicative of an erythropoietic response. In study 2, 79 patients received a single-dose infusion of 500 mg iron dextran. After intravenous iron, CHr rose within 48 hours, peaked at 96 hours, and then fell toward baseline. Patients who were iron deficient by standard measures (serum ferritin < 100 ng/mL or transferrin saturation less than 20%) had a greater and a sustained CHr response to intravenous iron dextran. A CHr less than 28 pg at baseline predicted functional iron deficiency, defined as a corrected reticulocyte increase of greater than 1% to iron dextran, more accurately than transferrin saturation, ferritin, or their combination. Eighty-two percent of individuals who were iron deficient at baseline responded to intravenous iron with an increase in CHr of greater than 2 pg. Sixty percent of patients who were iron sufficient by usual iron indices also responded to intravenous iron with a CHr rise of greater than 2 pg, suggesting that they were, in fact, functionally iron deficient despite "normal" conventional iron parameters. We conclude that CHr may be a more sensitive marker of functional iron deficiency in rHuEPO-treated hemodialysis patients than percent transferrin saturation and ferritin, particularly in those with "normal" conventional iron parameters.