7031 Background: AML in older adults is associated with poor outcomes. The Moffitt Cancer Center AML Database was used to evaluate a very large cohort of patients (pts) age ≥ 70 with untreated AML to identify key prognostic variables affecting outcome. Methods: Overall survival (OS): Kaplan-Meier method and was compared across groups using the log-rank test. Association between OS and predictors: Cox regression model. Impact of participation of initial clinical trial on OS: Propensity score with stratified log-rank test. A predictive model for 12 month OS was developed using multiple logistic regression with backward elimination method. Results: Nine hundred eighty (980) pts were identified. M/F(%): 66/34. Median age at diagnosis: 75.7 years (range 70 – 95.7 years). De novo/secondary (%): 43/57. Fifty two % of pts had prior hematologic disease (AHD). Baseline karyotype at AML diagnosis: adverse in 31% and non-adverse in 58%. Baseline ECOG PS: 0-1 in 79%; ≥2 in 19%. Median OS was 7.1 months (95% CI 6.4 – 7.9) for the entire cohort. In the univariable model, factors associated with inferior survival included: secondary AML (sAML) status, poor-risk karyotype, ECOG ≥2, non HMA therapy (including clinical trials), Charlson Comorbidity Index ≥3, older age, increased WBC, decreased platelets (plts), and decreased hemoglobin (hgb). Independent negative predictors for OS in the multivariate model included sAML, poor-risk karyotype, ECOG ≥2, non-HMA initial therapy, older age, increased WBC, decreased plts, and decreased hgb. Propensity score matching revealed no significant difference in OS amongst pts receiving initial treatment on a clinical trial (median 7.8 months, 95% CI 6.4 – 10.4) vs not (median 7.0 months, 95% CI 6 –7.9). A model to predict OS at 12 month was developed in a subset of 446 pts. Independent predictive variables included karyotype, ECOG PS, AML type (de novo vs sAML), age, and WBC, with AUC of 0.78, indicating strong discriminatory capacity. Conclusions: In this largest reported cohort of AML pts age ≥ 70, prognostic modeling identifies differences in longer-term survival with conventional therapies, discriminating the highest risk subsets. Decision modeling to further assist choice of optimal therapies for these pts is in progress.