Widely varied chemicals--including certain herbicides, plasticizers, drugs, and natural products--induce peroxisome proliferation in rodent liver and other tissues. This phenomenon is characterized by increases in the volume density and fatty acid oxidation of these organelles, which contain hydrogen peroxide and fatty acid oxidation systems important in lipid metabolism. Research showing that some peroxisome proliferating chemicals are nongenotoxic animal carcinogens stimulated interest in developing mode of action (MOA) information to understand and explain the human relevance of animal tumors associated with these chemicals. Studies have demonstrated that a nuclear hormone receptor implicated in energy homeostasis, designated peroxisome proliferator-activated receptor alpha (PPARalpha), is an obligatory factor in peroxisome proliferation in rodent hepatocytes. This report provides an in-depth analysis of the state of the science on several topics critical to evaluating the relationship between the MOA for PPARalpha agonists and the human relevance of related animal tumors. Topics include a review of existing tumor bioassay data, data from animal and human sources relating to the MOA for PPARalpha agonists in several different tissues, and case studies on the potential human relevance of the animal MOA data. The summary of existing bioassay data discloses substantial species differences in response to peroxisome proliferators in vivo, with rodents more responsive than primates. Among the rat and mouse strains tested, both males and females develop tumors in response to exposure to a wide range of chemicals including DEHP and other phthalates, chlorinated paraffins, chlorinated solvents such as trichloroethylene and perchloroethylene, and certain pesticides and hypolipidemic pharmaceuticals. MOA data from three different rodent tissues--rat and mouse liver, rat pancreas, and rat testis--lead to several different postulated MOAs, some beginning with PPARalpha activation as a causal first step. For example, studies in rodent liver identified seven "key events," including three "causal events"--activation of PPARalpha, perturbation of cell proliferation and apoptosis, and selective clonal expansion--and a series of associative events involving peroxisome proliferation, hepatocyte oxidative stress, and Kupffer-cell-mediated events. Similar in-depth analysis for rat Leydig-cell tumors (LCTs) posits one MOA that begins with PPARalpha activation in the liver, but two possible pathways, one secondary to liver induction and the other direct inhibition of testicular testosterone biosynthesis. For this tumor, both proposed pathways involve changes in the metabolism and quantity of related hormones and hormone precursors. Key events in the postulated MOA for the third tumor type, pancreatic acinar-cell tumors (PACTs) in rats, also begin with PPARalpha activation in the liver, followed by changes in bile synthesis and composition. Using the new human relevance framework (HRF) (see companion article), case studies involving PPARalpha-related tumors in each of these three tissues produced a range of outcomes, depending partly on the quality and quantity of MOA data available from laboratory animals and related information from human data sources.