574 Background: Omission of axillary surgery is appropriate in some patients with clinically node-negative (cN0), HR+ ESBC; however, there are no pre-operative tools to predict pathologic node positivity (pN+) in these women. We propose a clinically validated predictive model to inform treatment decisions regarding axillary evaluation. Methods: We constructed a cohort of adult women with ESBC (clinical T1/T2, N0, M0) diagnosed 2012-2016, who underwent lumpectomy or mastectomy and lymph node surgery without neoadjuvant therapy using the National Cancer Database breast cancer dataset. The dataset was non-randomly split into training (2012-2015) and testing (2016) for development and validation. Stepwise logistic regression was used to identify predictors of pathologic node positivity (pN0 vs pN+) in the training dataset. Potential predictors included: age, race, ethnicity, comorbidity score, histologic type, clinical T, ER positivity, PR positivity, HER2 positivity, and grade. Predictor variables required a bivariate p-value <0.30 to be entered into model, and an adjusted p-value <0.35 to stay in model. A partial score method was used to develop a lymph node prediction score (LNPS) by assigning a weighted value to each strong predictor variable (OR >1.5) and adding together the values for each included variable. LNPS was treated as a linear variable for prediction in validation dataset. Results: 423,068 women were included (2012-2015: 334,778; 2016: 88,290). Pathologic node positivity was 17% in 2012-2015 and 2016. All variables were included in the final stepwise model. Strong predictors were age, histologic type, clinical T, and grade. Scores ranged from 0-11. In the validation dataset, predicted pN+ by LNPS was very similar to actual pN+ (Table). A 1-point increase in LNPS was associated with a 3.3% increase in absolute risk of pN+. Conclusions: A novel lymph node prediction score can be used in HR+ cT1-T2 cN0 breast cancers to estimate the probability of pN+ and guide decisions regarding axillary surgical evaluation. [Table: see text]