The partial hepatectomy method, co-treatment method with mitogens and an in vivo/in vitro assay method have been reported as in vivo liver micronucleus (MN) assays. These methods have disadvantages with respect to widespread use as an in vivo assay, i.e. they are time consuming, labour intensive and there is the possibility of interaction with the mitogens used. Therefore, we have attempted to develop a new method to overcome these disadvantages. The assay as described herein utilises the autonomous proliferation of hepatocytes of young rats. Nine chemicals have been evaluated using this method thus far. We have also assessed the sensitivity and detectability according to the following methods. A liver MN assay was performed in two strains of young rats using one or two doses of 12 chemicals to investigate the inducibility of micronucleated hepatocytes. For some of the chemicals, a peripheral blood MN assay was performed concurrently in the same animals. The following chemicals were used: diethylnitrosamine (DEN), 2-acetylaminofluorene (2AAF), 2,4-diaminotoluene (2,4-DAT), quinoline, p-dimethylaminoazobenzene (DAB), dimethylnitrosamine (DMN), ethylmethanesulphonate, 5-fluorouracil, mitomycin C (MMC), 1,2-dimethylhydrazine.2HCl, cyclophosphamide and 2,4-dinitrotoluene (2,4-DNT). The rodent hepatocarcinogens, quinoline, DAB and DMN showed positive responses in previous assays. The results of the present assay revealed new positive responses for single doses of 2AAF, 2,4-DAT, MMC, 1,2-dimethylhydrazine.2HCl and 2,4-DNT. These chemicals are known rodent hepatocarcinogens, whose clastogenicity is believed to be related to the formation of reactive metabolites generated through enzymatic activation, or the chemicals act directly. Two doses of 2AAF and DMN appeared to be more effective than a single dose in terms of MN induction. Although there were quantitative differences in the incidences of MNs, both strains of rat (F344 and SD) responded positively after treatment with DEN, DMN, 2,4-DAT, DAB, quinoline and 2AAF, suggesting that both strains are appropriate for the assay. Based on these results, it is concluded that this technique could be effective for detecting chemical clastogenicity in hepatocytes in vivo.