The behavior of breast cancers and their response to different chemotherapy treatments depend on their phenotype which is to a large extent determined by gene expression programs within the cancer cell. We used gene expression values from 30 probes on affymetrix U133A chips to calculate a score that would predict if a breast cancer is likely to achieve pathological complete response to chemotherapy consisting of T/FAC. Overall design: Pre-treatment FNA from primary tumors or axillary lymph node metastases were obtained and RNA extracted and hybridized to afymetrix microarrays according to manufacturer protocol.