Background: Breast magnetic resonance imaging (MRI) use for surgical staging is increasing, though remains controversial. We aimed to evaluate the accuracy of MRI in surgical decision-making to determine if mastectomy prompted by MRI was appropriate. Methods: A single-centre observational study in Perth, Western Australia, with the inclusion of all preoperative and postoperative studies (e.g. involved margins after breast-conserving surgery) undergoing staging breast MRI from 1 January 2015 to 26 August 2019. A standard protocol using gadolinium contrast was used. The reference standard was postoperative histopathology or, for studies without additional surgery after MRI following breast-conserving surgery, the next and subsequent annual screening episodes. By reviewing the final histopathology, the medical case notes and multidisciplinary team decision process, we evaluated whether the reported MRI disease extent was accurate in prompting an appropriate upgrade to mastectomy. Outcomes are reported with descriptive statistics. Results: Of 130 cancers staged with MRI; seven were excluded as information was incomplete, 104 were performed preoperatively and 19 postoperatively. The majority (60%) staged lobular carcinoma (invasive 59%, in situ 1%) compared to ductal carcinoma (invasive 31%, in situ 8%). For preoperative MRI, half (54% – 56/104) underwent subsequent mastectomy. Of these, MRI prompted mastectomy in 45% (25/56), all appropriate for disease extent. In the postoperative staging group, two mastectomies were performed, one planned before imaging, the other prompted when MRI diagnosed residual disease and confirmed on histopathology. No false-negative staging MRI was identified. Conclusions: In our cohort, MRI prompted an upgrade to mastectomy in 21% (26/123), appropriate for cancer extent.