In the framework of computer-aided diagnosis of breast cancer, many systems were designed for the detection, the classification and/or the content-based mammogram retrieval (CBMR); in order to serve as a second source of decision for the radiologists. Nevertheless, to improve the final decision-making, the concept of multi-view information fusion (MVIF) was recently introduced. Indeed, this concept has been successfully applied in the context of breast cancer, since screening mammography provides two views for each breast: MedioLateral-Oblique (MLO) and CranioCaudal (CC) views. As these two views are complementary, MVIF methods widely proved their effectiveness. In this paper, we review the main methods that have been proposed for MVIF in the context of the detection (abnormality vs. non abnormality), the classification (normal vs. benign vs. malignant) and the content-based retrieval of mammograms. In fact, we classified detection based on MVIF methods into two main sub-classes, including ipsilateral analysis and bilateral analysis. Besides, classification based on MVIF methods were regrouped into two sub-classes, namely classification of breast masses based on MVIF and classification of breast microcalcifications based on MVIF. Lastly, CBMR based on MVIF methods were also classified into two sub-classes: early fusion-based MVIF-CBMR and late fusion-based MVIF-CBMR.