Recent low-dose carcinogenesis studies, including major group projects are reviewed. The prevailing paradigm is that carcinogens, particularly genotoxic compounds, have no threshold in exerting their potential for cancer induction. However, the nonthreshold hypothesis can be challenged for cancer risk assessment in humans. A recent very large-scale cooperative effort in Japan furthermore showed that the genotoxic hepatocarcinogen, 2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline, forms DNA adducts and 8-hydroxy-2'-deoxyguanosine at low doses, but does not induce glutathione S-transferase placental form (GST-P) positive foci as preneoplastic lesions in rat liver (< or = 10 ppm in diet). Moreover, very low doses of a N-nitroso compound. diethylnitrosamine (DEN), were also found not to significantly induce GST-P positive foci in rat liver (< or = 0.01 ppm in drinking water). Given the direct correlation between induction of the preneoplastic lesions in the short-term and carcinomas in the longer term with different carcinogens, the results imply a practical nonobserved effect level for hepatocarcinogenicity. Similar results were also observed with so-called nongenotoxic carcinogens such as phenobarbital (PB) and p,p-dichlorodiphenyltrichloroethane (DDT), which do not exert positive effects on lesion development at very low doses. Furthermore, experiments with application of PB and DDT after treatment with DEN indicate that at very low doses (< or = 2 ppm in diet), they may even inhibit the development of GST-P positive foci. The data reviewed provide evidence that preneoplastic foci in the liver can be employed as end-point lesions in place of tumors and that exposure to very low levels of carcinogens, typical of those found in the human environment, does not necessarily present as a risk factor.