Despite recent improvements in genotoxicity protocols, we have observed an increase in the occurrence of positive results, particularly in chromosomal aberration tests in vitro, yet very few of these are accompanied by positive responses in vivo. Thus, the positive results may not be biologically relevant either for rodents or humans in vivo, but how should we determine "biological relevance"? Chemicals that produce thresholded dose-responses may well not pose a genotoxic risk at low (relevant to human) exposures, but thresholds should not just be "seen"; there must be an explanation and understanding of the underlying mechanism. In addition to extremes of pH, ionic strength and osmolality, as have been identified previously, such mechanisms include indirect genotoxicity resulting from interaction with non-DNA targets, chemicals/metabolites which are inherently genotoxic but which, at low concentrations, are effectively conjugated and unable to form adducts, and production of specific metabolites under in vitro conditions that are not formed in rodents or humans in vivo. If such thresholded mechanisms can be identified at exposures which are well in excess of expected human exposure, then there may be a strong argument that the positive results are not biologically relevant.