Introduction: Treatment resistance, long latency, and high recurrence rates suggest that breast cancers arise from defective breast stem cells. Hypothesis: Within cancers, subpopulations of cells will demonstrate differences in stem/progenitor potential, HER2/neu amplification, and gene expression. Related cells will be found in normal breast tissue. Methods: ER-/PR-/HER2/neu + breast cancer cells were flow-sorted into subpopulations: (A) CD49f+ CD24-, (B) CD49f+CD24+, (C) CD49f CD24 -, and (D) CD49f-CD24+. Gel matrix cell invasion, fluorescence in situ hybridization (FISH) HER2/neu amplification, and qRT-PCR gene expression were measured in all groups. Cells from sorted groups were implanted into rat brains. Resultant tumors were analyzed by immunohistochemistry (IHC) and FISH. Normal breast tissue was examined by IHC. Results: Tumor development varied among sorted groups (25-75%), but was highest in group A. Tumor cells were mostly CD49f-CD24-, with variable fractions of other stem/progenitor cells. Tumors showed HER2/neu amplification, but fewer chromosome 17 per cell than inoculates. Group A tumors exhibited cells with normal chromosome 17 copy number and near normal HER2/neu amplification. Cell invasion was 61% higher in unsorted cells and 34-42% in sorted groups compared with controls. Sorted groups showed significantly different expression of development, proliferation, and invasion associated genes. In normal breast tissue, CD49f+ cells were identified in CD14+ CK19- basal epithelial layers of mammary glands; these were 95% CD24+ and 60% CD44+. Conclusions: Breast cancer stem/progenitor cell populations differ in tumor-initiating potential but are not solely responsible for metastasis. Cancer stem/progenitor cells are less polyploid than cancer cells in general and may not be HER2/neu amplified. In normal breast tissue, breast stem/progenitor cell-like populations are present.