OBJECTIVES: To compare the demographic, clinical and treatment characteristics of post-menopausal women with advanced ER+/HER2- breast cancer (BC) treated with aromatase inhibitors (AI) who experienced 0 or  ≥ 1 AI failure (AIF).  METHODS: Women ≥ 55 years old, newly diagnosed with metastatic ER+/HER2- BC (index) were identified from the 2006-2010 Thomson Reuters MarketScan databases.  Patients in the 6-month pre- or variable post-index periods treated with endocrine (ET: tamoxifen, fulvestrant) or AI (anastrozole, letrozole, or exemestane) therapy (ER+) and not with trastuzumab or lapatinib (HER2-), with no pre-index diagnosis of primary cancer other than BC, and post-index treatment with ≥ 1 AI were retained.  AIF was defined post-index as a switch to an alternative AI, ET, or chemotherapy, or AI discontinuation with no further BC treatment. RESULTS:  Among 4274 ER+/HER2- BC patients studied, 61% had ≥ 1 AIF (80% had 1 and 20% had 2+ AIFs).    There was no difference in pre-index AI use (54.4% no AIF, 51.8% AIF; p=0.093).   At index, AIF patients were more likely to be Medicare-eligible (57% vs. 51%) with liver (7% vs. 4%), lung (10% vs. 8%), bone (56% vs. 48%), and brain (7% vs. 5%) metastases, all p<0.03.   Mean follow-up days was shorter for AIF patients (486 vs. 522, p=0.006).  First line AI and ET treatments were respectively 95% and 5% for AIF and 97% and 3% for no AIF patients.  The most common first line therapy was anastrazole (49% no AIF 42% AIF).  Overall, with each subsequent line of therapy the proportion of AIF patients taking fulvestrant increased.  In addition, of those with AIF, 35% (n=905) switched to chemotherapy prior to study end.  CONCLUSIONS: The majority of ER+/HER2- metastatic breast cancer patients treated with AIs will fail at least 1 line of therapy.  Prior AI treatment did not appear associated with future AI failure.