OBJECTIVES:  Magnetic resonance imaging (MRI) is a sensitive method of breast imaging. The goal was to compare the effectiveness of MRI vs. mammography for post-treatment surveillance in patients undergoing breast-conserving therapy for breast cancer. METHODS: This cohort study used the Texas Cancer Registry linked Medicare claims data from 2004-2012. The study cohort included older women (age>65 years) diagnosed with ductal carcinoma in situ or SEER Historic stage 0, 1 and 2 breast adenocarcinoma who underwent breast-conserving surgery+radiation therapy. The primary independent variable was use of MRI +/- mammography vs. mammography alone after definitive therapy. Outcomes included time to local recurrence, number of total biopsies, and number of negative biopsies defined as biopsy not followed by breast cancer diagnosis and treatment. Independent variables included patient demographics, comorbidities and clinical characteristics such as cancer stage and tumor size. Unadjusted and adjusted Cox proportional hazards regression models were constructed to evaluate the comparative effectiveness of MRI versus mammography on time to detection of recurrence. Unadjusted and adjusted zero-inflated Poisson regressions were constructed to compare the incidence of total and negative biopsies with MRI vs. mammography. RESULTS:  Of 7,693 patients, 6,360 (82.7%) used mammography only, 483 (6.3%) used MRI and mammography and 850 (11.1%) had no post-treatment surveillance breast imaging. The use of MRI increased over time from 0.5% in 2004 to 17.7% in 2012 (p<0.001). In adjusted analysis, MRI was associated with higher detection of recurrence (HR 1.56; 95% CI, 1.15-2.13), but also associated with higher number of total biopsies (β=0.95[0.08]) and negative biopsies (β=1.01[0.09]). CONCLUSIONS:  MRI use for post-treatment surveillance increased in Texas in the last decade. The use of MRI resulted in greater detection of recurrence. Use of MRI was also associated with an increased number of total and false positive biopsies; follow-up was not long enough to evaluate if the early detection improved survival.