Identifying novel and known genes that are di erentially expressed in breast cancer has important implications in understanding the biology of breast tumorigenesis and developing new diagnostic and therapeutic agents. In this study we have combined two powerful technologies, PCRbased cDNA subtraction and cDNA microarray, as a high throughput methodology designed to identify cDNA clones that are breast tumor-and tissue-speci®c and are overexpressed in breast tumors. Approximately 2000 cDNA clones generated from the subtracted breast tumor library were arrayed on the microarray chips. The arrayed target cDNAs were then hybridized with 30 pairs of¯uorescent-labeled cDNA probes generated from breast tumors and normal tissues to determine the tissue distribution and tumor speci®city. cDNA clones showing overexpression in breast tumors by microarray were further analysed by DNA sequencing, GenBank and EST database searches, and quantitative real time PCR. We identi®ed several known genes, including mammaglobin, cytokeratin 19, ®bronectin, and hair-speci®c type II keratin, which have previously been shown to be overexpressed in breast tumors and may play an important role in the malignance of breast. We also discovered B726P which appears to be an isoform of NY-BR-1, a breast tissue-speci®c gene. Two additional clones discovered, B709P and GABA A receptor p subunit, were not previously described for their overexpression pro®le in breast tumors. Thus, combining PCRbased cDNA subtraction and cDNA microarray allowed for an e cient way to identify and validate genes with elevated mRNA expression levels in breast cancer that may potentially be involved in breast cancer progression. These di erentially expressed genes may be of potential utility as therapeutic and diagnostic targets for breast cancer.