PURPOSE: The most important prognostic factors of breast cancer are the tumor size and lymph node metastases. The accuracy of breast MRI in estimating tumor size has been investiated. Also, the role of breast MRI in diagnosing breast cancer has been studied since 1970s. METHODS: Between June 1997 and July 1998, 60 female patients with clinically and/or mammographically, sonographically suspicious breast carcinoma underwent surgery (surgical biopsy, lumpectomy, or mastectomy). With Gd-DTPA (ga dolinium-diethylenetriamine pentaacetic acid), contrast-enhanced MR imaging of the breast was per formed in all patients. we analyzed the specificity and the sensitivity of breast MRI compared to breast sonography and the mammography. We correlated the preoperative measurements of breast tumor size obtained using mammography, ultrasonography, and MRI with the values obtained from pathologic examination. RESULTS: The average age of the patients was 44.5 years and 48 patients (80%) had a pal pable mass. On histopathologic examination, 40 of 60 patients (66.7%) had malignant lesions. The specificities and the sensitivities of breast MRI, sonography and mammography were 73.7% and 85.0%, 70.0% and 92.9%, and 66.7% and 66.7%, respectively. MRI measurements demonstrated the highest correlation coefficient (r=0.895), with the lowest residual standard deviation (1.323) calculated in relation to the regression line. The correlation coefficient between sonographic tumor size and pathologic tumor size was 0.807 and the residual standard deviation was 2.194. CONCLUSION: The specificity and the sensitivity of the breast MRI were higher than those of mammography, but there was no difference between MRI and sonography. Breast MRI yielded the most accurate determination of breast cancer size in this series. As a result, MRI was useful for the diagnosis and the measurement of the tumor size.