Purpose: Preoperative tumor size is associated with clinical stage, treatment plan and even survival rate of patient. We investigated the accuracy of tumor size estimation between magnetic resonance imaging (MRI) and ultrasonography (US) findings, comparing these with pathologic tumor size in the diagnosis of early breast carcinoma. Methods: Between 2011 and 2016, 136 patients with early breast cancer were analyzed and their tumor size on US and MRI findings were compared with their pathologic tumor size retrospectively. The background parenchymal enhancement of MRI was categorized as minimal, mild, moderate, and extreme. The patients who underwent neoadjuvant chemotherapy, had positive resection margins, underwent excisional biopsy for cancer diagnosis, and had non-mass lesions on MRI scans, were excluded. Results: In all, 83.1% of the cases showed concordance between MRI findings and pathologic tumor size within 0.5cm. MRI overestimated the findings by 10.3% and underestimated them by 6.6%; 78.7% showed concordance between US findings and pathologic tumor size within 0.5cm. US overestimated the findings by 5.9% and underestimated them by 15.4%. The tumor size on MRI (r=0.87) showed a stronger correlation to the pathologic tumor size than that on US (r=0.64) in early breast cancer patients. US had a tendency to underestimate the tumor size. The degree of breast parenchyma did not affect the accuracy of the measurement of preoperative tumor size. Conclusion MRI is relatively more accurate than US for assessing preoperative tumor size in breast cancer patients. US tends to underestimate tumor size.