During the development of new industrial and pharmaceutical chemicals, it is necessary to determine whether they are potential carcinogens. However, there are no short-term tests available for nongenotoxic carcinogens that do not damage DNA yet cause tumours in rodent bioassays. The peroxisome proliferators (PPs) constitute a diverse class of nongenotoxic carcinogens that include chemicals of therapeutic, industrial and environmental importance such as hypolipidaemic fibrate drugs, clingwrap/medical tubing plasticizers and certain pesticides and solvents. PPs induce DNA synthesis and suppress apoptosis in rat and mouse hepatocytes, leading to tumour formation. In addition to altering hepatocyte growth and survival, PPs cause peroxisome proliferation and the induction of enzymes of the beta-oxidation pathway. PPs mediate their biological responses in rodents via activation of the nuclear hormone receptor PPARalpha (peroxisome proliferator activated receptor alpha) which regulates expression of the genes associated with response to PPs. The mechanisms through which normally quiescent hepatocytes are recruited into the cell cycle currently remain obscure. However, it is probable that expression of hepatic cytokines by hepatic macrophages (Kupffer cells) may be involved. In common with other classes of nongenotoxic carcinogen, there are remarkable species differences in response to PPs; humans respond to the fibrate hypolipidaemic PPs via a reduction in serum cholesterol but appear refractory to the adverse effects of PPs such as hepatic peroxisome proliferation, DNA synthesis and tumour formation. The molecular basis of the observed species differences in response to PPs is unclear at present, but recent data support a quantitative hypothesis wherein PPARalpha expression levels are sufficient in humans to mediate hypolipidaemia, but too low for transcriptional regulation of the full battery of genes associated with the adverse effects seen in rodents such as peroxisome proliferation, liver enlargement and tumours. A more detailed understanding of the mechanisms through which these chemicals cause tumours in rodents and how humans may differ will assist in extrapolation of rodent data to human risk assessment.