Appropriate use of magnetic resonance imaging (MRI) in elderly breast cancer (BC) patients remains unclear; we sought to identify the indications and implications of MRI use in our elderly BC population. Women 70 years of age or older at first BC diagnosis with an MRI performed at our institution either perioperatively or in follow-up were included from a prospectively maintained database from 2000 to 2010. Univariate logistic regression was used to test associations with disease identified by MRI only (additional ipsilateral, contralateral, or new cancer) following perioperative MRI. 305 BCs were imaged in 286 patients. 133 were imaged with MRI in the perioperative setting alone, 88 had only follow-up MRIs after BC treatment, and 65 had both. Indications for perioperative MRI include: extent of disease evaluation (181; 91%); occult primary (10; 5%); high-risk screening (5; 3%); and abnormal physical exam with negative conventional imaging (2; 1%). Disease identified by MRI only for occult primary cases was 4/10 (40%; 95% confidence interval: 12.2-73.8%) and 14/181 (7.7%; 95% confidence interval: 4.3-12.6%) for perioperative MRIs performed for extent of disease evaluation. Analysis of imaging and tumor characteristics failed to find significant predictors of disease identified by MRI only. A total of 369 post-treatment follow-up MRIs were performed in 148 patients with a median of 2 MRIs per patient (range 1-8), with seven cases of disease identified by MRI only (1.9%; 95% confidence interval: 0.8-3.9%). MRI had the greatest benefit in women presenting with an occult primary cancer and minimal additional benefit in elderly patients with BC undergoing MRI imaging for extent of disease evaluation or in post-treatment surveillance.