The purpose of this study was to compare the ability of magnetic resonance imaging (MRI) and mammography to determine the presence and extent of ductal carcinoma in situ (DCIS). Retrospective review of medical records of women who underwent MRI and mammographic examination during a 23-month period revealed 39 sites of pure DCIS in 33 breasts of 32 women. No invasive or microinvasive tumor was found. Women ranged in age from 34 to 79 years (mean age 53 years). In these 33 breasts, both MRI and mammography were done before surgery. Reports and images of mammography and MRI were reviewed to determine if each study was positive for the presence of single or multiple sites of DCIS and the imaging patterns associated with these sites. Of 33 breasts involved, DCIS was discovered by MRI alone in 21 (64%), by both MRI and mammography in 8 (24%), and by mammography alone in 1 (3%); in 3 breasts (9%), DCIS was found at mastectomy without findings on mammography or MRI. MRI had significantly higher sensitivity than mammography for DCIS detection (29/33=88% versus 9/33=27%, p<0.00001). Multiple sites of disease were present in five breasts; these were better demonstrated with MRI in three, mammography in one, and equally by both in one. The predominant enhancement pattern of DCIS on MRI was linear/ductal in 18 of 29 breasts (62%); mammography found calcifications associated with DCIS in 8 of 9 (89%). The nuclear grade of DCIS found with MRI and mammography was similar; size of lesions was larger on MRI; breast density did not impact results. In this study, MRI was significantly more sensitive than mammography in DCIS detection. In women with known or suspected DCIS, MRI may have an important role to play in assessing the extent of disease in the breast.