The investigation of micronuclei in mitogenic stimulated hepatocytes in vitro is a quite new area of research. Nevertheless, a relatively large database comprising more than 40 tested compounds of various classes has been generated up to now. This paper reviews the available data for the in vitro rat hepatocyte micronucleus assay, showing a sensitivity of this assay in identifying mutagens and genotoxic liver carcinogens of about 85%. Additionally, all of the tested non-carcinogens gave negative results. The use of primary hepatocytes instead of permanently dividing mammalian cell lines for the investigation of micronucleus induction has several advantages. (1) The broad spectrum of metabolizing enzymes expressed in primary hepatocytes ensures an adequate activation of most xenobiotics. (2) No transfer of activated metabolites via the culture medium is necessary in this system, since the metabolizing cells are the target cells themselves. (3) Whilst in experiments with permanently dividing cells the use of S9-mix restricts the treatment period with the test compounds to 2-6 h in the hepatocyte micronucleus assay continuous treatment of up to 48 h is possible. Investigations with the pyrrolizidine alkaloids retrorsine, monocrotaline and isatidine, strong mutagens and liver carcinogens, clearly showed that at least for isatidine a prolonged exposure period is essential to detect its mutagenic potential. This compound gave positive results in rat hepatocytes but not in V79-cells/S9-mix cultures. (4) The results obtained with the hepatocyte micronucleus assay are in good agreement with the genotoxic profiles of most of the compounds tested. Only three polycyclic aromatic hydrocarbons led to 'false-negative' results, since they strongly inhibited hepatocyte proliferation and thereby prevented micronucleus formation. (5) Hepatocytes are target cells of special interest when compounds are investigated which act specifically in the liver. Especially for hepatocarcinogens classified as non-genotoxins in standard genotoxicity tests or for chemicals showing DNA-repair induction in hepatocytes but no mutagenicity in standard tests, the hepatocyte micronucleus assay can contribute to clarify the situation. (6) The rat hepatocyte micronucleus assay can be performed easily and without great efforts in parallel to the in vitro hepatocyte DNA repair test (UDS-test), using the same hepatocyte batches. (7) Similar to the two versions of the UDS-test, the hepatocyte micronucleus assay can be performed following an in vivo-in vitro protocol. In order to further validate the hepatocyte micronucleus assay, as a next step controlled interlaboratory studies should be initiated.