Cancer has long been viewed as a heterogeneous population of cells. While the great majority of cells that make up tumors are destined to differentiate, albeit aberrantly, and eventually stop dividing, only a minority population of cells, termed cancer stem cells, possess extensive self-renewal capability and can recapitulate tumor pathophysiology in an immune-compromised animal model. Tumor-initiating cells have been identified and isolated in a variety of cancers of the blood, breast, central nervous system, pancreas, skin, head and neck, colon, and prostate. In this review we present scientific evidence supporting the cancer stem cell model of tumor progression, and discuss the experimental and therapeutic implications. The concept of cancer stem cells may have profound implications for our understanding of tumor biology and for the design of novel treatments targeted toward these cells. Current therapeutic strategies include targeting the cancer stem cell as well as its microenvironmental niche. We present an interesting, novel strategy that takes into account the reactive oxygen species status in cancer stem cells and how it might serve as a method for eradicating these cells in tumor growth.