The purpose of the workshop "Do Peroxisome Proliferating Compounds Pose a Hepatocarcinogenic Hazard to Humans?" was to provide a review of the current state of the science on the relationship between peroxisome proliferation and hepatocarcinogenesis. There has been much debate regarding the mechanism by which peroxisome proliferators may induce liver tumors in rats and mice and whether these events occur in humans. A primary goal of the workshop was to determine where consensus might be reached regarding the interpretation of these data relative to the assessment of potential human risks. A core set of biochemical and cellular events has been identified in the rodent strains that are susceptible to the hepatocarcinogenic effects of peroxisome proliferators, including peroxisome proliferation, increases in fatty acyl-CoA oxidase levels, microsomal fatty acid oxidation, excess production of hydrogen peroxide, increases in rates of cell proliferation, and expression and activation of the alpha subtype of the peroxisome proliferator-activated receptor (PPAR-alpha). Such effects have not been identified clinically in liver biopsies from humans exposed to peroxisome proliferators or in in vitro studies with human hepatocytes, although PPAR-alpha is expressed at a very low level in human liver. Consensus was reached regarding the significant intermediary roles of cell proliferation and PPAR-alpha receptor expression and activation in tumor formation. Information considered necessary for characterizing a compound as a peroxisome proliferating hepatocarcinogen include hepatomegaly, enhanced cell proliferation, and an increase in hepatic acyl-CoA oxidase and/or palmitoyl-CoA oxidation levels. Given the lack of genotoxic potential of most peroxisome proliferating agents, and since humans appear likely to be refractive or insensitive to the tumorigenic response, risk assessments based on tumor data may not be appropriate. However, nontumor data on intermediate endpoints would provide appropriate toxicological endpoints to determine a point of departure such as the LED10 or NOAEL which would be the basis for a margin-of-exposure (MOE) risk assessment approach. Pertinent factors to be considered in the MOE evaluation would include the slope of the dose-response curve at the point of departure, the background exposure levels, and variability in the human response.