In this review we have evaluated the relationship between peroxisome proliferation and hepatocarcinogenesis. To do so, we identified all chemicals known to produce peroxisome proliferation and selected those for which there are data (on peroxisome proliferation and hepatocarcinogenesis) which meet certain criteria chosen to facilitate comparison of these phenomena. The summarised data and definition of the methodology used has been collected in appendices. These comparisons enabled us to evaluate the relationship between these phenomena using reliable data. As there is a good correlation between them, we further explored the mechanisms of action that have been proposed (direct genotoxic activity, production of hydrogen peroxide, cell proliferation and receptor activation). The relationship between these events in other species, including humans, was also reviewed and finally an overview of the assessment of human hazard is presented in section IX. Some of the first chemicals which were shown to produce peroxisome proliferation were also hepatocarcinogens whose carcinogenicity could not be readily explained by genotoxic activity. This raised the suggestion that the unusual phenomenon of peroxisome proliferation was intricately linked to the carcinogenic activity of these agents. Three questions have exercised the attention of regulatory, industrial and academic toxicology since then; are chemicals which elicit peroxisome proliferation in the liver actually a coherent class of chemical carcinogens?; does the early biological phenomenon of peroxisome proliferation have real predictive value for and mechanistic association with rodent carcinogenesis?; and what hazard/risk do these agents pose to humans that may be exposed to them? Whether peroxisome proliferators are indeed a discrete class of rodent carcinogens would appear to be the single, most important question. If so, then the assumptions and procedures relevant to human hazard and risk assessment should be applied to the class and should be essentially generic; if not, each chemical should be considered independently. Our critical analysis of the published data for over 70 agents which have been shown to possess intrinsic ability to induce peroxisome proliferation in the livers of rodents has led to the conclusion that there exists a strong correlation between peroxisome proliferation as n early effect in the liver and hepatocarcinogenicity in chronic exposure studies. An almost perfect correlation was observed between the induction of peroxisomes in the rodent liver and the eventual appearance of tumours following chronic exposure The few exceptions to this were largely explainable (section II).(ABSTRACT TRUNCATED AT 400 WORDS)