BACKGROUND AIMS: NK-92, a permanent natural killer (NK) cell line, shows cytotoxicity against a variety of tumors and has been tested in a phase I trial. We tested the toxicity of NK-92 and chemotherapy drugs against the stem cell capacity of the acute leukemia cell line, KG1. While the chromium-release assay is the most common method for assessing cytotoxicity of immune effectors, and flow cytometry is increasingly used, the relationship of either assay to clonogenic readouts remains unknown. METHODS: KG1 was assessed for stem cell frequency by serial dilution, single-cell sorting and colony growth in methylcellulose. KG1 was sorted into CD34(+) CD38(+) and CD34(+) CD38⁻ populations and recultured in liquid medium or methylcellulose to determine the proliferative capacity of each fraction. Cytotoxicity of NK-92, daunorubicin and cytarabine against KG1 was measured using the chromium-release assay, flow cytometry and clonogenic assays. RESULTS: The culture-initiating cell frequency of whole KG1 was between 1 in 100 to 1000 by serial dilution and single-cell sorting. Although a rare (1-3%) CD34(+) CD38⁻ population could be demonstrated in KG1, both fractions had equivalent proliferative capacity. The cumulative flow cytotoxicity assay was more sensitive than the chromium-release assay in detecting target cell killing. At a 10:1 ratio NK-92 eliminated the clonogenic capacity of KG1, which was not predicted by the chromium-release assay. CONCLUSIONS: Clonogenic assays provide a more sensitive means of assessing the effect of a cytotoxic agent against putative cancer stem cells within cell lines, provided that they grow well in liquid culture medium or methylcellulose.