For a long time, biological properties of breast cancer have been classified histologically in view of histological type and grades. Although these histological classifications were effective for the description of pathology report and for the prediction of patients' prognosis, they were not very specific for the prediction of response of the cancer cells to systemic drug therapies. In 2000, Perou et al. proposed the molecular classification of breast cancer based on molecular profiles identified by using DNA microarray technique. They classified invasive ductal/lobular carcinomas and ductal carcinoma in situ into five subtypes: luminal A, luminal B, HER2 (c-erbB-2)-enriched, basal-like, and normal breast-like. These subtypes were not only correlated with patient prognosis but also useful for choosing effective drug therapies to patients. Because the assay system of molecular classification was expensive and was not established for diagnostic use, a surrogate ‘intrinsic subtype’ classification has come to be used widely. In the ‘intrinsic subtype’ classification, invasive breast cancers were categorized into four groups according not only to the expression status of hormone receptors (HR) and HER2 but also to Ki67 labeling index. If Ki67 labeling index is unreliable, substitution by histological/nuclear grade is allowed. Therefore, today's pragmatic ‘intrinsic subtype’ classification is based on the markers for indication of hormonal therapy and anti-HER2 therapy and a marker for tumor cell proliferation. Luminal A is defined as HR(+)/HER2(−)/Ki67-low, luminal B as either HR(+)/HER2(−)/Ki67-high or HR(+)/HER2(+), HER2-enriched as HR(−)/HER2(+), and triple-negative breast cancer (TNBC) as HR(−)/HER2(−). Problems remain in the classification in the cut-off value of the HR status; the method and cut-off value on Ki67 measurement; retesting with in situ hybridization of immunohistochemically HER2-equivocal cases; and subclassification of TNBC according to biological properties and sensitivity to therapies. Detailed histopathological and/or immunohistochemical examinations based on further research data might aid for much more useful classification toward individualized therapies.