Genome-wide expression profiling has expedited our molecular understanding of the different subtypes of breast cancers, as well as defined the differences among genes expressed in primary tumors and their metastases. Laser-capture microdissection (LCM) coupled to gene expression analysis allows us to understand how specific cell types contribute to the total cancer gene expression signature. Expression profiling was used to define genes that contribute to breast cancer spread into and/or growth within draining lymph nodes (LN). Whole tumor xenografts and their matched whole LN metastases were compared to LCM captured cancer cells from the same tumors and matched LN metastases. One-thousand nine-hundred thirty genes were identified by the whole organ method alone, and 1,281 genes by the LCM method alone. However, less than 1% (30 genes) of genes that changed between tumors and LN metastases were common to both methods. Several of these genes have previously been implicated in cancer aggressiveness. Our data show that whole-organ and LCM based gene expression profiling yield distinctly different lists of metastasis-promoting genes. Contamination of the tumor cells, and cross reactivity of mouse RNA to human-specific chips may explain these differences, and suggests that LCM-derived data may be more accurate.