The aim of the study was to evaluate the role of hypochromic erythrocytes (HYPO%) compared to "traditional" and novel markers of iron status and erythropoiesis in recognizing iron-restricted erythropoiesis (IRE) and predicting response to erythropoietin (rHuEPO) in anemic patients with myeloma and lymphoma. Forty-one newly diagnosed patients who received epoetin-beta at a subcutaneous weekly dose of 30,000 IU for 6 weeks were studied. Response to rHuEPO was observed in 27 patients (65.8%). Twelve non-responders received, additionally, 200 mg of IV iron sucrose, weekly, for 4 weeks. Evaluation of markers was performed at baseline and on weeks 1, 2 and 6 for all patients and also on weeks 7-10 for non-responders to rHuEPO. Baseline HYPO%, at a cut-off value of <5%, and an increment in reticulocyte absolute number (RETICS-AB) >or= 50,000/microl and reticulocyte hematocrit (RETICS-Hct) >or= 50%, between baseline and week 2, were independent predictive factors for response to rHuEPO. We found that these markers had superior predictive value for response to rHuEPO than four widely used predictive models. Furthermore, a baseline HYPO% count of above 5% proved superior over serum ferritin < 100 ng/ml and transferrin saturation < 20% in recognizing IRE. In conclusion, baseline HYPO% either alone or in combination with RETICS-AB or RETICS-Hct after 2 weeks of rHuEPO treatment could be reliably used in predicting response to rHuEPO. Additionally, HYPO% has proved a reliable marker for recognizing IRE before rHuEPO treatment and, thus, could be used for identifying patients who will benefit from IV iron supplementation.