We have used complementary DNA expression array and subtractive suppressive hybridization techniques to identify molecular alterations associated with the progression of breast cancer. Arrays representing up to 40,000 different transcripts were profiled against a large panel of clinical samples carefully selected by stage, histological type and clinical outcome. Samples included 6 normal breast epithelial cell, 8 ductal carcinoma in situ, 5 infiltrating lobular carcinoma and 28 infiltrating ductal carcinoma tumor samples. In addition, 11 cDNA subtraction libraries were generated from various comparisons of tumor versus normal samples and from comparisons of aggressive versus indolent tumor samples. Analysis of over 50,000 successful sequencing lanes from the subtraction libraries identified up to 6,000 unique transcripts expressed in breast cancer versus normal breast epithelium or in aggressive breast tumors versus indolent tumors. These 6,000 transcripts were assembled on custom arrays and have been hybridized to the same panel of breast samples. Analysis of the library and array data has led to the identification of several genes up- and down-regulated in breast cancer. A subset of these genes may represent new markers for breast cancer screening and prognosis as well as potential targets for future drug therapies.