Breast carcinomas represent an important health problem. Understanding the development of breast cancer from precursor is critical for clinical treatment and prevention, however little is known about the molecular events involved in the progression to cancer. The advent of gene expression microarray technology provides a new powerful tool to assist in the determination of diagnosis, prognosis and treatment. In this paper, we present the recent DNA microarray studies that describe how gene expression profiling is being used to classify specimens of breast carcinomas based on molecular properties of the tumor and to identify gene expression patterns related to clinical outcome. In present, data are available that show that gene expression profiles can be used to distinguish cell type-specific gene clusters (stromal, epithelial, mesenchymal and proliferation status) and to classify breast tumors as basal-like, luminal-like, ERBB2 overexpressing and normal breast-like. Profiles associated with good prognosis and poor prognosis of young axillary node negative patients have been identified. The microarray technology will become in the near future a molecular complement to histopathology and immnuhistochemistry.