Background: Older patients with human epidermal growth factor 2-positive (HER2+) metastatic breast cancer(mBC) are underrepresented in clinical trials. We aim to assess the overall survival (OS) and breast cancer-specific survival of elderly women with de novo HER2+/hormone receptor-negative (HR-) mBC in a real-world setting. Methods: Elderly women with HER2+/HR- mBC treated with chemotherapy and/or HER2-targeted agents and with continuous Medicare Part A, B, and D coverage 1-year before diagnosis were identified from the SEER-MEDICARE database 2012-2016. Patients were retrospectively followed from metastatic diagnosis until death, disenrollment from Medicare A, B, or D, or end of the observation period. Patients' year and month of diagnosis and death were retrieved from SEER. Death dates were verified with Medicare records reported by the Social Security Administration (SSA). For all-cause deaths, Kaplan-Meier analysis was used to estimate overall survival. The cumulative incidence competing risk (CICR) method based on cumulative incidence function (CIF) was used to estimate breast cancer-specific death incidence. Results: Seventy-three patients (mean age at diagnosis, 75.0±7.7 years) met the inclusion criteria. Among them, 56 were treated with trastuzumab ± pertuzumab /chemotherapy as first-line treatment, and 17 were treated with chemotherapy only. The median time to initiate trastuzumab-based treatment from diagnosis was 2.5 months, and the longest trastuzumab treatment length was over 44 months. The median follow-up for OS was 13 months. One patient developed stomach cancer 6 months after breast cancer diagnosis. In Kaplan-Meier analysis, censoring or not censoring this patient after second cancer development resulted in a median OS of 19 months (95% CI, 9-24 months) and 18 months (95% CI, 9-22 months). The OS at the end of 46 months was approximately 25%. Five patients died from other causes, including lung cancer, cerebrovascular diseases, aortic aneurysm and dissection, pneumonia and influenza, and heart diseases during treatment. Considering these competing risks, 50% (95% CI, 36%-64%) of patients specifically died from breast cancer between 21 and 22 months, estimated by the CICR method. Conclusions: Our study observed a shorter OS among HER2+/HR- mBC elderly patients in clinical practice than the OS of 40.8 and 56.5 months among younger patients in the CLEOPATRA trial, suggesting that age is an important prognostic factor for breast cancer survival. The presence of second cancer and other competing risks led to overestimating the probabilities of breast cancer-specific death and resulted in a shorter OS using the Kaplan-Meier method. The CICR method is more relevant to estimate the breast-cancer-specific death incidence.