The distribution of genotoxic factors in various organs of mice treated orally with methylazoxymethanol-beta-D-glycoside (cycasin) was investigated using the DNA-repair host mediated assay. Indicator of genotoxic activity was a pair of streptomycin dependent Escherichia coli strains differing vastly in DNA repair capacity; uvrB/recA vs. uvr+/rec+. The animal-mediated assays were performed by injecting mixtures of the two strains i.v. and orally into mice, which were subsequently treated with the test chemical and from which the differential survival of the indicator bacteria present in several organs was determined. The same strains and selection procedures were also used for assessing the DNA-damaging activity in vitro. In the animal-mediated assays in which cycasin was applied orally, significant effects were observed at doses of 100 and 500 mg/kg body weight. The organ distribution of genotoxic factors in the host animal was as follows: the highest genotoxic activity was observed in the liver, followed by intestine and stomach; a clear effect was also observed in the kidneys and, to a lower extent, in the blood stream and in the lungs at the highest dose administered (500 mg/kg body weight). Under in vitro conditions a marginal genotoxic effect was observed even in the absence of liver homogenate, indicating that the test compound is possible activated (hydrolysed) by the E. coli cells. Therefore the genotoxic activity of cycasin observed in the gastrointestinal tract was not unexpected, since the substance was applied orally, thereby exposing the indicator bacteria in these organs to high levels of unmetabolised compound, especially in the stomach. In the intestine members of the microbial flora probably contribute to the metabolic activation of the test compound. The occurrence of genotoxic factors remote from the gastrointestinal tract shows that the present compound or active metabolites thereof penetrate through the intestinal barrier. The extraordinarily high genotoxic activity observed in the liver suggests that the compound is additionally activated in this organ. In compliance with previous in vitro findings this second activation step might lead to the formation of the highly reactive aldehydic form of methylazoxymethanol (MAMAL) mediated by dehydrogenases. Comparison with carcinogenicity studies indicates a good correlation between the distribution of genotoxic effects as determined in the present studies and the localisation of tumors in various organs of rodents treated with cycasin.