Azacitidine has recently been approved for the treatment of International Prognostic Scoring System (IPSS) intermediate-2 and high risk myelodysplastic syndromes (MDS). In these subsets of disease it is able to induce 50–60% of responses, including 10–20% complete remissions and partial remissions [1]. Results from the randomized international trial AZA001 showed that azacitidine treatment was associated with improved survival if compared with conventional treatments, and appeared to be the best strategy in the setting of patients with high risk MDS [2]. Prognostic factors associated with the probability of response and overall survival remain yet unclear. Recently, the French group of Itzykson and colleagues [3] reported a large analysis of prognostic features at baseline in 282 patients with MDS treated with azacitidine, and found that performance status 2, presence of peripheral blasts, transfusion dependency 4 units/8 weeks and intermediate and poor risk karyotype independently predicted shorter overall survival (OS). The authors created a prognostic score able to stratify patients into three groups (low, intermediate and high risk) and identify differences in OS [3]. The score was then validated, but in a series of selected patients from the AZA001 trial. The aim of the present study was to validate this simple score in a series of non-selected patients with MDS treated with azacitidine from a single institution. For this analysis, 60 patients were recruited: there were 44 males and 16 females, median age 69 years (range 44–83). According to World Health Organization (WHO) criteria [4], 35 patients were classified as having refractory anemia with excess of blasts type 2 (RAEB-2), 13 patients as having RAEB-1, eight patients as having refractory cytopenia with multilineage dysplasia (RCMD) and four patients as having chronic myelomonocytic leukemia type 2 (CMML-2). IPSS stratification revealed 51 patients as intermediate-2 risk and nine patients as high risk. Cytogenetic analysis displayed aberrations in 30 patients and a normal karyotype in the other 30; the most common karyotypic abnormality was monosomy 7 (10 patients, 20%), isolated or associated with complex changes. Seven patients had treatment-related MDS (t-MDS): five patients had received chemotherapy for non-Hodgkin lymphoma, one patient had received chemotherapy for chronic lymphocytic leukemia and one patient for a solid tumor. Median follow-up was 15 months. All patients were treated with the same dose of azacitidine (75 mg/m 2) with the same schedule (5 2 2). The median number of cycles received was 6 (range 2–21). After a median of four cycles, according to 2006 International Working Group (IWG) criteria [5], 15 patients (25%) achieved a complete response (CR), three patients (5%) a partial remission (PR), 20 patients (33%) a hematologic improvement (HI) and seven (12%) patients had stable disease. Fifteen patients (25%) experienced a progression of disease. Statistical analysis referred to cumulative incidence of response compared by Gray test. We applied the French score to our series of patients and identified 12 patients as having low risk, 38 patients as having intermediate risk and 10 patients as having high risk disease. We found a statistically significant difference in terms of median OS in these three subgroups of patients: 21 months for low risk, 15 months for intermediate risk and 11 months for high risk (p 0.001). According to IWG 2006 criteria [5] and according to the French stratification, we found that this score was also able to identify patients who could reach a CR with azacitidine. This was achieved after four cycles in 6/12 low risk patients (50%), in 9 …