Micronuclei in exfoliated epithelial cells are widely used as biomarkers of cancer risk in humans. To elucidate the effect of different staining procedures on the outcome of such investigation, we conducted a study in which the micronuclei frequencies in oral mucosa cells of heavy smokers (n = 20) and nonsmokers (n = 10) were evaluated with nonspecific (Giemsa, May-Grü nwald-Giemsa) and DNA-specific (4 ¶,6-diamidino-2-phenylindole, Feulgen, acridine orange) stains, whereas with Giemsa-based stains, the frequencies of micronuclei in smokers were significantly (4-to 5-fold) higher in the smokers group, no significant increase was observed with any of the DNA-specific stains. Furthermore, the evaluation of cells of the two study groups with Feulgen stain showed that oral mucosa cells from smokers had significantly increased levels of nuclear anomalies other than micronuclei.These anomalies are consequences of cell injury found in epithelial cells and are paralleled by formation of keratin bodies in the cytoplasm that resemble micronuclei. Correlation analyses showed that micronuclei frequencies scored in Giemsa-stained slides correlated significantly with karyorrhexis, karyolysis, condensed chromatin, and binucleates, whereas no such correlations were found with DNA-specific stains. These findings indicate that nuclear anomalies (and possibly keratin bodies) may be misinterpreted as micronuclei with nonspecific DNA stains and lead to false-positive results in studies with cells of epithelial origin. Furthermore, our results show that exposure of oral mucosa cells to genotoxic carcinogens contained in tobacco smoke does not lead to induction of micronuclei in these cells. (Cancer Epidemiol Biomarkers Prev 2006;15(10):1835 -40)