Abstract The concept of precision medicine, where therapy is personalized to unique disease and host characteristics, is currently revolutionizing treatment strategies in oncology. A key point in this concept is the fact that several so-called targeted drugs may only be effective when tumors exhibit characteristic molecular features. This implies that companion diagnostic tests should be performed before treatment. There is evidence suggesting that the first step in predictive pathology is represented by morphology. For instance, histological types, such as tubular and cribriform carcinomas, define patients who may not need any treatments other than surgical excision. Neoadjuvant studies have shown that patients affected by lobular carcinomas are not likely to have any beneficial effects from chemotherapy. The second step in prediction is represented by immunophenotyping. If the immunohistochemical evaluation of four markers (estrogen and progesterone receptors, HER2 and Ki67) remains the best practice for breast cancer predictive pathology, molecular pathology has certainly reshaped the way we approach breast cancer diagnosis. The aim of this review is to discuss current knowledge in predictive pathology for the management of breast cancer patients, focusing on the benefits and drawbacks of traditional tools and of novel improvements of molecular biology.