Often results from toxicological studies using rodent models cannot be directly extrapolated to probable effects in human beings. In order to examine the genotoxic potential of chemicals in human liver cells, a human hepatocyte DNA repair assay has been defined. Procedures were optimized to prepare primary cultures of human hepatocytes from discarded surgical material. On eight different occasions human hepatocyte cultures of sufficient viability to measure DNA repair were successfully prepared by collagenase perfusion techniques. The cells were allowed to attach to plastic or collagen substrata for periods of 1.5 to 24 h and subsequently incubated with [3H]thymidine and test chemicals for periods of 18 to 24 h. Chemically induced DNA repair, measured as unscheduled DNA synthesis, was quantitated autoradiographically. The following compounds were tested: 2-acetylaminofluorene, aflatoxin B1, 2-aminobenzyl alcohol, aniline, benzo(a)pyrene, carbon tetrachloride, chloroform, 2,4-diaminotoluene, 2,6-diaminotoluene, di(2-ethylhexyl)phthalate, dimethylnitrosamine, 1,6-dinitropyrene, 2,4-dinitrotoluene, 2,6-dinitrotoluene, methyl chloride, 5-methylchrysene, mono(2-ethylhexyl)phthalate, 2-methyl-2-P-(1,2,3,4-tetrahydro-1-naphthyl)phenoxypropionic acid (nafenopin), beta-naphthylamine, nitrobenzene, 2-nitrobenzyl alcohol, 2-nitrotoluene, 2,3,7,8-tetrachlorodibenzo-p-dioxin, unleaded gasoline, and 4-chloro-6-(2,3-xylidino)-2-pyrimidinylthioacetic acid (Wy-14,643). In only one of eight cases did some of the chemicals generally regarded as genotoxic fail to give a positive response. For purposes of comparison, all test chemicals were evaluated in the in vitro rat hepatocyte DNA repair assay. Individual-to-individual variation in the DNA repair response was far greater for the human cultures than for cultures derived from rats. For only three chemicals was there a qualitative difference in the response between the rodent and the human cells; beta-naphthylamine was positive in the rat but in none of the human cultures examined, whereas the opposite was seen for 2,6-diaminotoluene and 5-methylchrysene. Clofibric acid, mono(2-ethylhexyl)phthalate, and Wy-14,643 induced enzymes indicative of peroxisomal proliferation in primary rat hepatocyte cultures, but not in two human hepatocyte cultures. These results indicate that, in general, the in vitro rat hepatocyte DNA repair assay is a valid model for predicting potential genotoxic effects in human beings. However, rodent hepatocytes may not be appropriate for assessing the potential of chemicals to elicit nongenotoxic effects in human beings such as the induction of hepatocyte peroxisomal proliferation.