Dear Sir, Human recombinant erythropoietin (Hu-Epo) has enabled significant advances in the management of anaemic patients with myelodysplastic syndrome (MDS)1. Hu-Epo provides erythroid precursors protection from apoptosis and stimulates the residual normal haematopoiesis2,3, although possible effects on abnormal clones have also been reported4,5. An excessive erythroid response may be observed in some patients who are usually managed by an appropriate dose reduction of the Hu-Epo and/or by an extended interval of administration. However, we recently observed an erythroid response that was maintained for up to several months despite therapy discontinuation. This extremely rare phenomenon occurred in a 76-year old man who was admitted to hospital in July 2011 because of severe malaise and fatigue. On admission, besides the severe macrocytic anaemia (haemoglobin [Hb] =5.2 g/dL) with decreased reticulocyte count requiring the transfusion of six units of red blood cells; the neutrophil count (950/μL) and platelet count (123,000/μL) were also slightly reduced. The bone marrow was hypercellular with erythrodysplasia and 2% of blasts but no dysgranulopoietic and/or dysmegakaryocytic features were observed. All other causes of anaemia were carefully ruled out; the serum level of endogenous erythropoietin was in the normal range. Therefore, a diagnosis of MDS, subtype refractory anaemia, was made. Conventional cytogenetic and fluorescence in situ hybridization analyses identified a 20q deletion in 80% of metaphases. According to the International Prognostic Scoring System and the World Health Organization organisation-based Prognostic Scoring System, the MDS risk was classified as Intermediate-1 and very low risk, respectively. The patient received epoetin-α 40,000 IU twice a week and was followed up regularly. After 2 weeks, a major erythroid response was achieved so the dose of epoetin-α was reduced to 40,000 IU once a week. After 4 weeks, a complete haematological recovery was recorded: in particular, the patient’s Hb level was 12.5 g/dL and he had normal platelet and neutrophil counts, so the interval of administration of epoetin-α was extended to 2 weeks. After 8 weeks the haemoglobin level was 15.4 g/dL and the epoetin-α was withdrawn. Thereafter, he maintained near-normal Hb levels, ranging from 12.4 and 13.5 g/dL, without any therapeutic intervention. At present, 11 months after discontinuation of epoetin-α, the patient’s Hb level is 13.2 g/dL, although he continues to show the same bone marrow findings detected at disease onset, such as erythrodysplasia and the original clonal cytogenetic alteration in 50% of metaphases. In daily clinical practice it is very rare that patients responding to Hu-Epo have a prolonged normalisation of peripheral blood counts, without any further therapy, once treatment has been discontinued because of an excessive therapeutic response. To the best of our knowledge, only one case of MDS with a similar outcome after treatment with Hu-Epo and human granulocyte-macrophage colony-stimulating factor has been reported so far5. A possible interpretation of our findings is that the Hu-Epo may have enhanced erythropoieis both through maturation of MDS progenitor cells as well as through the reduction of karyotypically abnormal clones5 and the expansion of a non-clonal erythroid matrix2,3. The cytogenetic mosaicism in this case allows us to postulate that the therapy with Hu-Epo expanded the polyclonal erythropoiesis at the expense of the very low malignant clonal erythropoiesis. In conclusion, in our low-risk MDS patient a short course of Hu-Epo at high doses appears to have exerted, through unknown mechanisms, some durable effects on the processes of haematopoiesis and/or apoptosis restoring normal erythropoiesis in spite of the discontinuation of epoetin-α.