Background In the setting where determining extent of residual disease is key for surgical planning after neoadjuvant chemotherapy (NAC), we evaluate the reliability of MRI in predicting pathologic complete response (pCR) of the breast primary and axillary nodes after NAC. Study Design Patients who had MRI before and after NAC between June 2014 and August 2015 were identified in a prospective database after IRB approval. Post-NAC MRI of the breast and axillary nodes was correlated with residual disease on final pathology. Pathologic complete response was defined as absence of invasive and in situ disease. Results We analyzed 129 breast cancers. Median patient age was 50.8 years (range 27.2 to 80.6 years). Tumors were human epidermal growth factor receptor 2 amplified in 52 of 129 (40%), estrogen receptor-positive/human epidermal growth factor receptor 2-negative in 45 of 129 (35%), and triple negative in 32 of 129 (25%), with respective pCR rates of 50%, 9%, and 31%. Median tumor size pre- and post-NAC MRI were 4.1 cm and 1.45 cm, respectively. Magnetic resonance imaging had a positive predictive value of 63.4% (26 of 41) and negative predictive value of 84.1% (74 of 88) for in-breast pCR. Axillary nodes were abnormal on pre-NAC MRI in 97 patients; 65 had biopsy-confirmed metastases. The nodes normalized on post-NAC MRI in 33 of 65 (51%); axillary pCR was present in 22 of 33 (67%). In 32 patients with proven nodal metastases and abnormal nodes on post-NAC MRI, 11 achieved axillary pCR. In 32 patients with normal nodes on pre- and post-NAC MRI, 6 (19%) had metastasis on final pathology. Conclusions Radiologic complete response by MRI does not predict pCR with adequate accuracy to replace pathologic evaluation of the breast tumor and axillary nodes.