The selection process for chemicals tested in the rodent carcinogenicity bioassay has been biased toward chemicals suspected of potential carcinogenicity. Results from carcinogenicity bioassays of 400 chemicals tested by the National Cancer Institute/National Toxicology Program (NCI/NTP) were analyzed to determine the dependence of positive results on chemical selection criteria: those suspected of being carcinogenic and those selected based on large volumes produced and widespread exposures. Of these chemicals, 210 (52%) induced carcinogenicity in at least one organ of one sex of one species of the four sex/species groups typically used by NCI/NTP. Only 92 of the 400 chemicals (23%) were positive in two species and thus by international criteria are considered likely to pose a carcinogenic hazard to humans. A total of 267 chemicals (67%) were selected as suspect carcinogens, and 187 (68%) of these were carcinogenic. Suspect chemicals account for 86% of chemicals with at least one positive result and account for 90% of chemicals considered positive in two species. The International Agency for Research on Cancer (IARC) lists only 5 of the 400 chemicals as carcinogenic to humans (group 1) and 10 as probably carcinogenic to humans (group 2A). The majority (80%) of the 133 chemicals selected only on production/exposure considerations were not carcinogenic in animals, even when tested at the maximum tolerated (or minimally toxic) dose. Only 9 (6.8%) were positive in two species, and none is listed in IARC groups 1 or 2A. Thus, on the basis of our analyses we predict that less than 5-10% of the 75,000 chemicals in commercial use might be reasonably anticipated to be carcinogenic to humans.