Background: Reliable methods for the isolation of an enriched cancer stem cell (CSC) population in colorectal cancer (CRC) may provide new approaches for the identification of therapeutic targets. There is no consensus regarding the best methods to isolate a CRC tumor cell population enriched for CSCs. The goal of this study was to determine the utility of various CSC markers in both established cell lines and freshly isolated patient-derived tumor xenografts in order to validate methods of CSC isolation. Methods: Established human CRC cell lines (HCT116, HT29, SW480) and three freshly isolated CRC cell lines were studied. Freshly isolated cell lines were generated by first injecting CRC cells derived from clinical specimens into nude mice. Xenografted tumors were resected and FAC-sorted to isolate human cells. These cells were FAC-sorted for CD133, CD44, and ALDH activity. Tumorsphere formation and in vivo dilutional and serial tumorigenicity studies were done to validate methods for CSC enrichment. Results: Using established human CRC cell lines, none of the markers studied could be used to enrich for the CSC population as assessed by sphere formation or tumor initiating capacity; this suggests that cells grown in culture for long periods of time lose their hierarchy. In the freshly isolated CRC cells, CD133 and CD44 were unable to enrich for CSC-ness. In contrast, freshly isolated CRC cells with high ALDH activity formed spheres whereas cells with low ALDH activity did not. In validation studies, freshly isolated cells from patient-derived xenografts with high ALDH activity demonstrated enhanced tumorigenic capacity in murine models. Furthermore, the tumorigenicity of ALDH high cells was maintained after serial passage of cells harvested from tumors generated by ALDH high cells, while ALDH low cells did not form tumors. Conclusions: CRC cells from patient-derived tumor xenografts can be used to study CSC properties using ALDH activity as a marker. In contrast, CD44 and CD133 are not reliable markers for enriching for the CSC population from established cell lines or patient xenografts. Established CRC cell lines should not be used to study CSC biology.