Transfusion dependency seems to have a major prognostic impact in patients with myelodysplastic syndrome (MDS) (Malcovati L et al. J Clin Oncol2007;25:3503). Preliminary data also suggest that the development of iron overload could influence outcome (Malcovati L et al. J Clin Oncol2005;23:7594 and Garcia-Manero G et al. Leukemia2008;22:538), but small numbers have precluded a meaningful analysis of the prognostic value of this characteristic. The main aim of this study was to evaluate the independent prognostic value of transfusion dependency (as defined in WHO-based Prognostic Scoring System [WPSS]) and iron overload (defined as serum ferritin level >1,000 ng/mL) in a large series of 2,994 patients (median age, 74 yr) with de novo MDS according to FAB criteria (2,107 MDS according to WHO criteria). Complete transfusional history was available in 2,241 patients (835 transfusion dependent [TD] at diagnosis, 526 TD during follow-up, and 880 non-TD) and serum ferritin levels in 1,634. Karyotyping was successfully performed in 2,074 patients, who could then be classified by the International Prognostic Scoring System (IPSS) as low (861 patients), intermediate-1 (748), intermediate-2 (311), and high-risk (154). The numbers of patients in the five risk categories defined by the WPSS (available for 1,228 patients) were 257 (21%) in very low, 385 (31%) in low, 217 (18%) in intermediate, 271 (22%) in high, and 98 (8%) in very high, closely similar to those reported in the original WPSS series. Actuarial curves of overall survival (OS) and risk of evolution to acute myeloblastic leukemia (AML) were built by Kaplan-Meier method and differences between curves compared with log-rank tests. Multivariate analyses of OS and risk of evolution to AML were performed by Cox proportional hazards regression method, with development of transfusion dependency and iron overload entered as time-dependent covariates. Other variables included in the prognostic factor analyses were age, gender, hemoglobin level, absolute WBC, PMN, and platelet counts, proportion of blasts in blood and marrow, percentage of dysplastic features in the three different hematopoietic cell lines, cytogenetics according to IPSS cytogenetic risk subgroups, FAB and WHO classifications, ferritin, beta-2 microglobulin, erythropoietin, and LDH levels at diagnosis, and IPSS and WPSS risk categories. All the previous variables showed a statistically significant relationship with OS and/or AML risk on univariante analyses. Median OS for TD patients at diagnosis, TD patients during evolution, and non-TD patients was 19, 60, and 96 months, respectively (P