6503 Background: In higher-risk MDS, 80% of first responses by IWG 2000 criteria (hematologic improvement or better; HI+) to AZA are achieved with a median of 6 treatment (Tx) cycles (Silverman, 2008). In the AZA-001 trial, an overall survival (OS) benefit with AZA vs conventional care regimens (CCR) was observed with a median of 9 AZA Tx cycles (Fenaux, 2009); CR was not necessary for prolonged OS. These findings provide little guidance for management of pts with a best response of stable disease without HI (SD) after 6 AZA cycles. We evaluated the impact of SD on OS in AZA- and CCR-Tx pts in AZA-001. Methods: OS was estimated using Cox proportional hazard models stratified by FAB and IPSS with a factor for Tx. Time-varying covariates of SD and HI+, and terms for Tx-by-response interactions were added. The best model was chosen using the Akaike Information Criterion and statistical significance of the covariates. Landmark analyses of pt response at 3, 6, and 9 mo and estimating OS (Kaplan Meier) were use...