In summary, the extent of toxicity studies which can be carried out with bound residues of nitroimidazole drugs will be dictated by the availability and the amount of residues that can be produced. For evaluating their toxicity the use of the Salmonella assay, which has been proven to be sensitive for the parent drugs, is proposed. Furthermore, it is suggested that the toxicity of bound residues for in vivo effects should be investigated in cells of target tissues, such as the epithelial cells of the gastrointestinal tract. Nuclear aberration and sister chromatid exchange assays in this tissue would be good candidates for evaluation. It should be pointed out also that an examination of the literature for genotoxic effects of nitroimidazole drugs reveals an apparent contradiction, especially in the context of genotoxicity testing strategy discussed above. It appears first that these drugs are potent mutagens in vitro in microbial systems (Salmonella and yeast). However, they are not active in mammalian cells in vitro, in both mutation and chromosomal aberration assays, as well as in vivo, in assays such as the dominant lethal test and the micronucleus assay. Thus, it may be of interest to speculate that, although these drugs are in vitro mutagens, they may not be in vivo mutagens in mammals. Their tumorigenic effects, which have been detected at high doses in rodents, may, therefore, be due to other than genotoxic activity. To resolve this conflict, reassessment of genotoxicity of these drugs in vivo would be a worthwhile pursuit. In addition, such seeming discrepancies would also argue strongly for toxicity screening to be conducted in a battery of complementary short-term in vitro and in vivo tests. This would help insure the likelihood of detection of at least some meaningful biological activity, which would serve to flag chemicals warranting further testing.