Toxicogenomics is a relatively new discipline of toxicology. Microarrays and bioinformatics tools are being used successfully to understand the effects of toxicants on in vivo and in vitro model systems, and to gain a better understanding of the relevance of in vitro models commonly used in toxicological studies. In this study, cDNA filter arrays were used to determine the basal expression patterns of human cultured primary hepatocytes from different male donors; compare the gene expression profile of HepG2 to that of primary hepatocytes; and analyze the effects of three genotoxic hepatocarcinogens; aflatoxin B(1) (AFB(1)), 2-acetylaminofluorene (2AAF), and dimethylnitrosamine (DMN), as well as one non-gentoxic hepatotoxin, acetaminophen (APAP) on gene expression in both in vitro systems. Real-time PCR was used to verify differential gene expression for selected genes. Of the approximately 31,000 genes screened, 3-6% were expressed in primary hepatocytes cultured on matrigel for 16 h. Of these genes, 867 were expressed in cultured hepatocytes from all donors. HepG2 cells expressed about 98% of the genes detectable in cultured primary hepatocytes, however, 31% of the HepG2 transcriptome was unique to the cell line. A number of these genes are expressed in human liver but expression is apparently lost during culture. There was considerable variability in the response to chemical carcinogen exposure in primary hepatocytes from different donors. The transcription factors, E2F1 and ID1 mRNA were increased three-fold and six-fold (P < 0.05, P < 0.01), respectively, in AFB(1) treated primary human hepatocytes but were not altered in HepG2. ID1 expression was also increased by dimethylnitrosamine, acetylaminofluorene and acetaminophen in both primary hepatocytes and HepG2. Identification of genes that are expressed in primary hepatocytes from most donors, as well as those genes with variable expression, will aid in understanding the variability in human reactions to drugs and chemicals. This study suggests that identification of biomarkers of exposure to some chemicals may be possible in the human through microarray analysis, despite the variability in responses.