Abstract Background: Sex steroid hormone signaling is critical in the progression of breast cancers, though the role of androgens remains largely unspecified. Evidence from cancer cell line models suggests that androgen receptor (AR) signaling inhibits cell proliferation in estrogen receptor positive (ER+) cancers and stimulates growth of ER- tumors. This hypothesis is supported by population-based studies of ER+ cancers, however the effect of AR in ER- cancers remains unclear, as population studies have had limited sample sizes and short durations of follow-up. Methods: We evaluated the association between AR expression and breast cancer survival among 4,792 pre- and post-menopausal women with invasive breast cancer in the Nurses' Health Study (NHS) and Nurses' Health Study II (NHSII) cohorts. The NHS and NHSII were established in 1976 and 1989, respectively, with the enrollment of U.S. registered nurses between the ages of 30 and 55 (NHS) or 25 and 42 (NHSII). Tumor tissue was obtained from breast cancer cases diagnosed between 1976 and 2006 (NHS) or 1989 and 2006 (NHSII). Tissue microarrays were constructed containing tumor tissue and AR and ER expression were measured by immunohistochemistry and scored manually by pathologists with ≥1% expression considered positive for AR and ER. The association between AR expression and survival was assessed through construction of Kaplan-Meier survival curves and through use of Cox proportional hazards models to estimate hazard ratios (HR) and 95% confidence intervals adjusted for patient, tumor, and treatment covariates. Results: Over follow-up through 2014 (median 15.3 years), a total of 1,666 deaths and 813 deaths due to breast cancer occurred. We observed that AR expression was associated with improved breast cancer-specific survival overall (p<0.001 from log-rank test) and among ER+ tumors (p<0.001), while there was a non-significant deleterious effect of AR expression among ER- tumors (p=0.076). These findings were robust in Cox proportional hazards models that accounted for patient, tumor, and treatment covariates. Overall, AR expression was associated with a 16% reduction in the breast cancer-specific mortality rate (HR=0.84, 95% CI 0.70-1.01). This association was more prominent among ER+ tumors, where breast cancer mortality was 31% lower for women who co-expressed AR (HR=0.69, 95% CI 0.55-0.88). In contrast, AR expression led to a 33% increase in breast cancer mortality among ER- tumors (HR=1.33, 95% CI 1.01-1.75), and a 44% increase in the subset of triple negative ER-/PR-/HER2- tumors (HR=1.44, 95% CI 1.01-2.07). Similar associations were observed when evaluating the outcome of overall survival and when using an alternative cut point of 10% for AR expression. We found no significant interactions between AR expression and treatment covariates. Conclusions: AR expression was found to be a robust predictor of breast cancer survival with the direction of the association dependent on tumor ER co-expression. These findings support the hypothesis that AR signaling pathways may be potential therapeutic targets using AR signaling agonists for ER+ cancers and AR signaling antagonists for ER- cancers. Citation Format: Kevin H. Kensler, Elizabeth M. Poole, Laura C. Collins, Andrew H. Beck, Bernard A. Rosner, A. Heather Eliassen, Susan E. Hankinson, Myles Brown, Rulla M. Tamimi. Androgen receptor expression and breast cancer survival in the Nurses' Health Study cohorts [abstract]. In: Proceedings of the AACR International Conference: New Frontiers in Cancer Research; 2017 Jan 18-22; Cape Town, South Africa. Philadelphia (PA): AACR; Cancer Res 2017;77(22 Suppl):Abstract nr A05.