Objectives: To evaluate the contribution of magnetic resonance imaging (MRI) to predict nipple involvement in breast carcinoma. Methodology: A total of 170 mastectomy specimens from patients affected by ductal carcinoma in situ (DCIS) (n = 13) or invasive ductal carcinoma (stages I, II, or IIIA) (n = 157) were examined. All breasts were evaluated preoperatively using 1.5 Tesla MRI. The nipples of the surgical specimens were evaluated in histological sections with a thickness of 3µm searching for DCIS and invasive carcinoma. Radiologist blinded to the histopathological results reviwed all MRI exams in order to evaluate the following parameters: type of index lesion enhancement, size of the index lesion, enhancement extending from the index lesion to the nipple, distance between the index lesion and the nipple, nipple enhancement, thickening of the nipple, retraction of the nipple and size of the nipple compared to contralateral. Results: Among those parameters investigated, only the enhancement extending from the index lesion to the nipple and the retraction of the nipple were statistically significant (p <0.001 and 0.010, respectively) in multivariate analyses. The negative predictive value of the association of both variables mentioned was 89.5%. Conclusion: This result allows us to conclude that the probability of finding the nipple free of neoplastic involvement is about 90% when there are no enhancement extending from the index lesion to the nipple and retraction of the nipple.