Sentinel lymph node biopsy (SLNB) has become an alternative to axillary lymph node dissection (ALND) despite the limited recidive long-term results. SLNB can not only reduce ALND morbidity but also provide ultrastadification with serial sectionning and immunohistochemistry analysis which increase the sensitivity of detection of sentinel node (SN) metastasis. Micrometastasis or isolated tumor cells are frequently discovered. However, their diagnostic and pronostic values are still subject to controversy. Most of large randomized trials have determined that double detection (colorimetric and isotopic) improved SN identification rate and decreased false negative rate; and that periareolar injection was equally effective, even superior than peritumoral injection with the major advantage of its simplicity in non palpable tumors. One of the unsolved problems of SLNB is to determine if its indications may be extended to larger tumors, to node sampling before or after neoadjuvant chemotherapy, or after previous lumpectomy or breast surgery, in case of palpable axillary node, and in case of multifocal tumor. Another challenge is to determine if complementary ALND in case of SLND metastasis is necessary, because 40 to 70% of non sentinel nodes (NSN) are tumor-free. Several predictive models (nomograms, scores, partitioning recursive models) have been developed to predict non-SN status in SN-positive patients. These models must be validated in independent cohorts to enable their use in routine.