Increases in the frequency of micronuclei (MN) in exposed cells can be used as a measure of genotoxicity. Hair dyes contain chemicals that are eliminated by urine and could be genotoxic to urothelial cells. To address this question, we evaluated whether hair dye use is associated with an increase in the frequency of MN in urothelial cells, and whether this association is modified by NAT1 (N-acetyltransferase 1), NAT2 (N-acetyltransferase 2) and GSTM1 (glutathione-S-transferase M1) genotypes. We included 92 women participating as controls in a bladder cancer case-control study in Spain. Of those, 72 had adequate number of cells to be included in the MN analysis. There were no significant differences in the mean MN frequency in women using hair dyes in the last month (9.88 MN/1000 cells), in comparison with the MN in unexposed women (9.50 MN/1000 cells). No statistically significant differences in MN frequency were observed by type of hair dye or color of the hair dye. Comparison of subjects in the highest quartile of MN frequency (> or = 12 MN/1000 cells) and those in the lowest quartile (< or = 4 MN/1000 cells) suggested an association between hair dye use and elevated MN frequency (OR 14.2 (95% CI 0.81-247.8; P=0.069)). None of the polymorphisms examined significantly modified association between hair dye use and frequency of MN. Findings of an increased frequency of MN in urothelial cells of hair dye users suggest a possible genotoxic effect of hair dye compounds and need confirmation in larger studies.