Our goal is to identify new molecular targets for drug design and improve understanding of the molecular basis of clinical behavior and therapeutic response of breast cancer (BC). Pure populations of BC cells were procured by laser capture microdissection (LCM) from deidentified tissue specimens. RNA from either LCM-procured cells or whole tissue sections was extracted, purified, and quantified by RT-qPCR using beta-actin for relative quantification. RNA was amplified, Cy5-labeled, and hybridized for microarray. Spectrophotometric and BioAnalyzer analyses evaluated aRNA yield, purity, and transcript length for gene microarray. Unsupervised and supervised methods selected 7,000 genes with significant variation. Expression profiles of BC cells were dominated by genes associated with estrogen receptor-alpha (ERalpha) status; over 3,000 genes were identified as differentially expressed between ERalpha+ and ERalpha(-) BC cells. Other prominent gene expression patterns divided ERalpha+ BCs into subgroups, which were associated with significantly different clinical outcomes (p < 0.01). While exploiting larger gene sets derived from LCM-cells and reports using whole tissues, a preliminary 14 gene subset was selected by UniGene Cluster analysis. Additionally, ERE-binding proteins (ERE-BP) were detected by EMSA, which were not recognized by ERa antibodies. Kaplan-Meier analysis indicated that patients with ERE-BP positive BCs had lower over-all survival than those with ERE-BP negative cancers. Collectively, these results will establish molecular signatures for assessing clinical features of BC and aid in the selection of molecular targets for drug development.