Background: Novel biomarkers are required to discern between breast tumors that should be targeted for treatment from those that would never become clinically apparent and/or life threatening for patients. Moreover, therapeutics that specifically target breast cancer (BC) cells with tumor-initiating capacity to prevent recurrence are an unmet need. We investigated the clinical importance of LGR5 in BC and ductal carcinoma in situ (DCIS) to explore LGR5 as a biomarker and a therapeutic target. Methods: We stained BC (n = 401) and DCIS (n = 119) tissue microarrays with an antibody against LGR5. We examined an LGR5 knockdown ER- cell line that was orthotopically transplanted and used for in vitro colony assays. We also determined the tumor-initiating role of Lgr5 in lineage-tracing experiments. Lastly, we transplanted ER- patient-derived xenografts into mice that were subsequently treated with a LGR5 antibody drug conjugate (anti-LGR5-ADC). Results: LGR5 expression correlated with small tumor size, lower grade, lymph node negativity, and ER-positivity. ER+ patients with LGR5high tumors rarely had recurrence, while high-grade ER- patients with LGR5high expression recurred and died due to BC more often. Intriguingly, all the DCIS patients who later died of BC had LGR5-positive tumors. Colony assays and xenograft experiments substantiated a role for LGR5 in ER- tumor initiation and subsequent growth, which was further validated by lineage-tracing experiments in ER- /triple-negative BC mouse models. Importantly, by utilizing LGR5high patient-derived xenografts, we showed that anti-LGR5-ADC should be considered as a therapeutic for high-grade ER- BC. Conclusion: LGR5 has distinct roles in ER- vs. ER+ BC with potential clinical applicability as a biomarker to identify patients in need of therapy and could serve as a therapeutic target for high-grade ER- BC.