1019 Background: Diverse mutational processes are active at different stages of cancer development and cause the mutational burden in cancer cells. These processes leave distinctive signatures, characterized by varying frequencies of nucleotide substitutions and their flanking base-pairs. Mutational signatures found in early stage breast cancer have been described, but less is known about the mutational processes in metastatic breast cancer (MBC). Methods: We performed whole exome sequencing (WES) on 209 metastatic tumor biopsies from patients with ER+ MBC. In a subset of patients, we obtained and performed WES on the matched primary biopsy (n = 62). Decomposition of mutational signatures was done by determining the linear combination of 30 previously reported mutational signatures (Alexandrov et al., 2013) that most accurately reconstructs the mutational profile of each tumor sample. For comparison, signature decomposition was also performed on 695 treatment-naïve primary ER+ breast cancers form The Cancer Genome Atlas (TCGA). Results: Cohort-level comparison of signature activity between our cohort of MBC biopsies versus the primary ER+ breast cancers from TCGA demonstrated enrichment of Signature 13 in MBC samples (q = 0.002), attributed to activity of the APOBEC family of cytidine deaminases and implicated in resistance to endocrine therapies. Activity of Signature 22, associated to activity of the nucleotide-excision repair mechanism, was depleted in MBC tumors relative to TCGA ER+ breast tumors (q = 0.02). Decomposition of mutational signatures at the clone level in the subset of patients with both metastatic and matched primary samples allowed us to infer signatures acquired in the metastatic setting. Among mutations in acquired clones, we found enrichment of APOBEC signatures (Sig. 2, q = 0.018; Sig. 13, q = 0.003). Conversely, activity of Signature 1, attributed to the normal aging process, was enriched in mutations shared between primary and MBC samples (q = 0.004). This suggests that mutations developed early in tumor evolution may be driven primarily by natural processes, while later mutations may reflect selective pressures imposed by therapies and cellular defects acquired during disease progression. Conclusions: The profile of somatic mutations in MBC differs from early stage breast cancer. APOBEC-related mutations arising late during tumor evolution are one possible mechanism for this difference. Further studying the associations between clinical features and therapies to mutational processes could inform design of clinical trials and management of MBC.