The present study was designed to investigate the impact of aluminum toxicity on the response to recombinant human erythropoietin (rHuEPO) therapy in hemodialysis patients, when iron deficiency has been corrected or excluded. We studied 39 patients on regular hemodialysis (20 males and 19 females; mean age 58.8 years), who were under maintenance rHuEPO treatment for at least 6 months, and who had stable hematocrit levels for more than 3 months. All patients had adequate iron stores and availability with serum ferritin >100 µg/l and iron saturation >25%. They were classified into two groups: 19 poor responders, who required subcutaneous rHuEPO doses >100 U/kg/week and failed to achieve the target hematocrit level of 30%, and 20 good responders, who needed doses of ≤100 U/kg/week to maintain the target level. Serum aluminum levels including basal (Al<sub>basal</sub>) and 44 h after desferrioxamine (DFO) infusion (Al<sub>post-DFO</sub>), intact parathyroid hormone, and inflammatory and hemolytic indices were examined in both groups. The results showed that the mean weekly rHuEPO doses were significantly lower and the mean hematocrit levels higher in the good responders than in the poor responders. Although the poor responders had markedly higher mean Al<sub>basal</sub> and Al<sub>post-DFO</sub> levels, no differences were observed in the other parameters between the two groups. Furthermore, the poor responders significantly had the greater increment in the serum aluminum levels after DFO infusion (ΔAl =Al<sub>post-DFO</sub> – Al<sub>basal</sub>). The mean corpuscular volume had a strong inverse correlation with ΔAl in the poor response group (r = –0.711, p < 0.001). We concluded that the post-DFO rise of serum aluminum can be used as a means of estimating tissue stores. Subclinical aluminum toxicity may exhibit an inhibitory effect on erythropoietic response to rHuEPO therapy.