Benzyl isothiocyanate (BITC) is contained in cruciferous plants which are part of the human diet. Numerous reports indicate that BITC prevents chemically induced cancer in laboratory animals and it has been postulated that BITC might also be chemoprotective in humans. On the other hand, evidence is accumulating that this compound is a potent genotoxin in mammalian cells by itself. To further elucidate the potential hazards of BITC, we investigated its genotoxic effects in different in vitro genotoxicity tests and in animal models. In in vitro experiments [differential DNA repair assay with Escherichia coli, micronucleus assay with human HepG2 cells and single cell gel electrophoresis (SCGE) assay with hepatocytes and gastrointestinal tract cells] pronounced dose-dependent genotoxic effects were found at low dose levels (≤ 5 μg/ml). In contrast, substantially weaker effects were obtained in in vivo experiments with laboratory rodents: in the differential DNA repair assay with E.coli cells, only moderate genotoxic effects were seen in indicator cells recovered from various organs of mice after treatment with high doses (between 90 and 270 mg/kg), while in SCGE assay with rats a change in the DNA migration pattern was seen at a dose level of 220 mg/kg body wt. These findings indicate that BITC is detoxified under in vitro test conditions. This assumption was supported by the results of in vitro experiments which showed that the genotoxic effects of BITC are markedly reduced by bovine serum albumin and human body fluids such as saliva and gastric juice. Additional experiments carried out on the mechanistic aspects of the genotoxicity of BITC showed that this compound causes formation of thiobarbituric acid-reactive substances in HepG2 cells and that its DNA damaging properties are diminished by α-tocopherol, vitamin C, sodium benzoate and β-carotene, indicating the possible involvement of free radicals in the genotoxicity of BITC. The doses of BITC required to cause measurable DNA damage in laboratory rodents exceeded by far the dietary exposure levels of humans, but are similar to those which were required to inhibit chemically induced cancer in earlier animal experiments.