Background: The greatest age disparity in early stage breast cancer (BC) outcomes is in young women with hormone receptor positive (HR+) BC. While differences in biology may play a role, understanding the role of non-persistence (early discontinuation) with endocrine therapy (ET) is critical given the demonstrated efficacy of ET in this population. Methods: As part of a prospective cohort that enrolled women with BC diagnosed (dx) at age ≤40 between 2006-2016, we identified women with HR+, Stage I-III BC. Socio-demographic and treatment information, fertility concerns and confidence with the ET treatment decision were assessed by survey within 1 yr of dx. Medical record review was used to ascertain stage and HR status. Women who initiated ET but did not report taking tamoxifen or an aromatase inhibitor at 3 yrs post-diagnosis (or last follow-up if <3 yrs) were classified as non-persistent. Chi-square tests were used to compare categorical variables between persisters and non-persisters and stepwise multivariable regression to evaluate predictors of non-persistence. Results: In 538 women who initiated ET, median age at dx was 36; 10% were non-persistent. Discontinuation of ET was more likely in those who were less confident with their ET decision compared to those who were more confident (25/179, 14% vs 18/263, 7%, p=0.01). A greater proportion of women concerned about fertility discontinued vs. women not concerned (29/213,14% vs 25/319, 8%, p=0.03), and fertility concerns were associated with non-persistence in multi-variable analyses (OR: 1.85, 95% CI 1.05-3.26, p=0.03). Age at dx, race, education, employment, financial comfort, marital status, parity, stage, chemotherapy and local therapy were not associated with non-persistence. Conclusions: A significant minority of women with HR+ BC discontinued ET within 3 yrs. The association between fertility concerns expressed soon after dx and non-persistence underscores a need to address psychosocial issues that can impact treatment decisions in young women. Strategies to reduce decisional conflict and increasing confidence with the choice to take ET, may influence persistence. Future work will evaluate the contribution of other factors (eg symptom burden) to non-persistence.