Human breast carcinomas represent a heterogeneous group of tumors diverse in behavior, outcome, and response to therapy. However, the current system of pathological classification does not take into account biologic determinants of prognosis. The purpose of this study was to classify and characterize breast carcinomas based on variations in protein expression patterns derived from immunohistochemical analyses on tissue microarrays (TMAs). Therefore, 11 TMAs representing 168 invasive breast carcinomas were constructed. Breast tumors were classified into four different subtypes depending on estrogen receptor (ER) and HER2 expression. Basal-type tumors expressed neither of these proteins and represented 7.6% of our series; basal-like HER2-overexpressing tumors did not express ER and represented 17.7%; luminal-type tumors expressed ER and represented 72.8% of this series (luminal A 56.3%, luminal B 16.5%). Moreover, we characterized each subtype based on P-cadherin (P-CD), p63, cytokeratin (CK)5, BCL2, and Ki67 expression. Basal-type tumors were mostly grade III, more frequently P-CD-, p63-, and CK5-positive, and had a high proliferation rate. Conversely, luminal-type tumors rarely expressed basal markers and had a low grade and proliferation rate. Basal-like HER2-overexpressing tumors showed a basal-type profile similar with a high grade and up-regulation of P-CD and CK5. With this study, we show that P-CD, p63, and CK5 are important molecular markers that can be used to distinguish a basal phenotype. In addition, we also demonstrate the usefulness of TMAs in breast carcinoma immunoprofiling.