The objective of this study was to assess the efficacy of MR mammography (MRM) in evaluating breast cancer extent in women with fatty or dense breasts, and its contribution to the therapeutic approach. The authors reviewed 97 carcinomas detected in 93 women (both symptomatic and from screening) that were classified in two groups according to breast density pattern. Mammography, ultrasound (US), and MRM were performed to evaluate size, extension of the in situ component, presence of multifocal/multicentric disease, and contralateral involvement. Results obtained on mammography plus US were balanced against MRM, considering pathologic analysis as the gold standard. For fatty breasts (n = 47), exact measurement was found on mammography plus US and on MRM alone in 70%, underestimation on mammography plus US 23.5% and on MRM 11% (P = 0.005). For dense breasts (n = 50), exact measurement was found on mammography plus US in 40% and on MRM alone 68%, underestimation on mammography plus US 52% and on MRM 10% (P = 0.005). Overall, good correlation (R > 0.71) was found between pathologic and clinical size with all imaging methods; nevertheless, when evaluating multifocal/multicentric disease, a poor correlation was observed between histologic assessment and mammography plus US (R = 0.52), but it was excellent with regard to MRM (R = 0.99). In fatty breasts, the combination of mammography and US allows for a precise assessment of tumoral extension. However, these results show that in dense breasts, MRM is superior to mammography plus US, suggesting that its systematic use in this group of patients is justifiable.