The current United States Environmental Protection Agency (EPA) classification of di(2-ethylhexyl)phthalate (DEHP) as a B2 "probable human" carcinogen is based on outdated information. New toxicology data and a considerable amount of new mechanistic evidence were used to reconsider the cancer classification of DEHP under EPA's proposed new cancer risk assessment guidelines. The total weight-of-evidence clearly indicates that DEHP is not genotoxic. In vivo administration of DEHP to rats and mice results in peroxisome proliferation in the liver, and there is strong evidence and scientific consensus that, in rodents, peroxisome proliferation is directly associated with the onset of liver cancer. Peroxisome proliferation is a transcription-mediated process that involves activation by the peroxisome proliferator of a nuclear receptor in rodent liver called the peroxisome proliferator-activated receptor (PPARalpha). The critical role of PPARalpha in peroxisomal proliferation and carcinogenicity in mice is clearly established by the lack of either response in mice genetically modified to remove the PPARalpha. Several mechanisms have been proposed to explain how, in rodents, peroxisome proliferation can lead to the formation of hepatocellular tumors. The general consensus of scientific opinion is that PPARalpha-induced mitogenesis and cell proliferation are probably the major mechanisms responsible for peroxisome proliferator-induced hepatocarcinogenesis in rodents. Oxidative stress appears to play a significant role in this increased cell proliferation. It triggers the release of TNFalpha by Kupffer cells, which in turn acts as a potent mitogen in hepatocytes. Rats and mice are uniquely responsive to the morphological, biochemical, and chronic carcinogenic effects of peroxisome proliferators, while guinea pigs, dogs, nonhuman primates, and humans are essentially nonresponsive or refractory; Syrian hamsters exhibit intermediate responsiveness. These differences are explained, in part, by marked interspecies variations in the expression of PPARalpha, with levels of expression in humans being only 1-10% of the levels found in rat and mouse liver. Recent studies of DEHP clearly indicate a nonlinear dose-response curve that strongly suggests the existence of a dose threshold below which tumors in rodents are not induced. Thus, the hepatocarcinogenic effects of DEHP in rodents result directly from the receptor-mediated, threshold-based mechanism of peroxisome proliferation, a well-understood process associated uniquely with rodents. Since humans are quite refractory to peroxisomal proliferation, even following exposure to potent proliferators such as hypolipidemic drugs, it is concluded that the hepatocarcinogenic response of rodents to DEHP is not relevant to human cancer risk at any anticipated exposure level. DEHP should be classified an unlikely human carcinogen with a margin of exposure (MOE) approach to risk assessment. The most appropriate and conservative point of reference for assessing MOEs should be 20 mg/kg/day, which is the mouse NOEL for peroxisome proliferation and increased liver weight. Exposure of the general human population to DEHP is approximately 30 microg/kg body wt/day, the major source being from residues in food. Higher exposures occur occupationally [up to about 700 microg/kg body wt/day (mainly by inhalation) based on current workplace standards] and through use of certain medical devices [e.g., up to 457 microg/kg body wt/day for hemodialysis patients (intravenous)], although these have little relevance because the routes of exposure bypass critical activation enzymes in the gastrointestinal tract.