A prospective study with 65 maintenance hemodialysis (MHD) patients on recombinant human erythropoietin (rHuEPO) therapy was conducted to assess the effect of iron balance on responsiveness. An attempt to define the predictors of erythropoietin (EPO) response and identify the specific causes of EPO resistance was undertaken in the present study. The treatment protocol consisted of two stages, the first was rHuEPO therapy for 6 months and the second was iron supplementation plus rHuEPO therapy in patients without response to EPO for the next 6 months. According to the hemoglobin (Hb) changes (increment exceeded 30% of baseline or did not exceed 15% of baseline for 3 consecutive months) and whether or not there was an achievement of target Hb level ( > 10.5 g/dl), all patients (n = 65) were divided into EPO-responsive (n = 20) and EPO-resistant (n = 45) groups. The EPO-resistant patients were then further stratified into iron-responsive (n = 29) and iron-irresponsive (n = 16) groups. Iron metabolism and red cell indices were analyzed prior to and following rHuEPO therapy and iron supplementation. We found the following: (1) only serum ferritin (SF) was a reliable discriminator between the EPO-responsive (SF > 300 µg/l) and EPO-resistant (SF < 300µg/l) groups; (2) similarly, transferrin saturation (TFS) 25% was quoted as the best cut-off value between the iron-responsive (TFS < 25%) and iron-irresponsive (TFS > 25%) groups; (3) EPO-resistant patients with TFS < 2 5 % regained proper EPO response (Hb before and 6 months after therapy: 7.8 ± 0.9 vs. 10.6 ± 0.8 g/dl, p < 0.01) and the mean TFS increased significantly (initial TFS and peak level after therapy: 18.9 ± 4.7 vs. 34.5 ± 10.8%, p < 0.01) following iron therapy; (4) the cumulative incidence of TFS < 25% elevated to 44.5% 6 months following initial rHuEPO therapy, and (5) there was a strong inverse relationship between initial TFS and the changes of Hb following iron therapy in EPO-resistant patients (r = -0.75, p < 0.0001). By means of early detection and correction of functional iron deficiency, the goal of increment of therapeutic efficacy and prevention from unnecessarily high doses of rHuEPO can be achieved. In summary, SF > 300 µg/l equates adequate or increased body iron stores and seems to exclude iron deficiency in MHD patients. On the contrary, SF < 300 µg/l appears to be an insufficient diagnostic threshold for iron deficiency. Once resistance occurs during the rHuEPO therapy, in our opinion only patients with TFS < 25% need iron supplementation. However, in patients with TFS > 25% and resistance to EPO, further investigations are necessary before increasing the dose of rHuEPO to explore other possible conditions, such as aluminum overload, severe hyperparathyroidism, occult blood loss or hemolysis, and episodes of infection or inflammatory processes.