Background: Although anemia of IPSS low and int-1 (LR) MDS initially responds to erythroid stimulating agents (ESA) in 40-50% of patients (pts), response is generally transient. Anemia recurrence with RBC cell transfusion dependency (RBC-TD) is an indicator of poor prognosis, even in absence of progression to higher risk MDS. AZA leads to RBC transfusion independence (RBC-TI) in 30–40% of LR-MDS (Lyons, JCO, 2009), but it has not been prospectively tested in LR-MDS patients resistant to ESA. It also remains unknown if the addition of ESA to AZA would be useful in such pts. Methods: In this randomized phase-II trial (GFMAzaEpo-2008-1 trial, NCT01015352), the Groupe Francophone des Myélodysplasies (GFM) compared AZA 75mg/m2/d for 5 days every 28 days for 6 cycles (AZA arm) to the same treatment plus EPO 60000U/week (AZA+EPO arm). Inclusion criteria were LR-MDS resistant to at least 12 weeks of ESA, with RBC-TD ≥ 4 RBC units in the previous 8 weeks. Responders in both arms were eligible for maintenance up to 12 monthly cycles, unless progression or loss of erythroid response occurred. The primary endpoint was RBC-TI (HI-E major using IWG 2000 criteria) after 6 courses. Results: Between 2009 and 2010, the 98 planned pts were included: M/F 65/28; median age 71y (41-84); 5 pts were excluded (one consent withdrawal, 2 early unrelated events and 2 with exclusion criteria). In the remaining 93 pts, IPSS risk was low in 69 and int-1 in 23 pts, with no imbalance between arms. Five pts received < 4 cycles and 16 received only 4 or 5 cycles, mainly due to toxicity or progression. RBC-TI after 6 courses was observed in 16.7% (8/48) in the AZA arm and 18 % (8/45) in the AZA+EPO arm (P=1), and in 19.5% (8/41) and 26 % (8/31) respectively, in the 72 pts who received ≥ 6 cycles (P=.57). Overall best response rate (at least HI-E minor using IWG 2000 criteria) was 35 and 33% in the AZA and AZA+EPO arms, respectively (P=0.99). Of note, only 9 pts in the AZA+EPO arm required at least one hospitalization for a SAE, compared to 22 pts in the AZA arm (P=0.015). Conclusions: Erythroid response to AZA, in LR MDS with demonstrated ESA resistance, appears lower than expected, with no improved outcome when combined with an ESA. The latter may however improve tolerance of AZA in LR MDS.