A physiologically based pharmacokinetic (PB-PK) model for CHCl3 has been used to prepare estimates of the probability that human populations exposed to low levels of CHCl3 will develop liver tumors similar to those seen in rodent bioassays. The PB-PK model for CHCl3 was based on a model reported earlier by Corley et al. (1990), but this model differed from that of Corley et al. in that it was also capable of describing a pharmacodynamic endpoint: induction of cytotoxicity in the liver of CHCl3-exposed animals produced by reactive metabolites of CHCl3. Pharmacodynamic descriptions in this model were derived from experimental measurements of cell replication ([3H]thymidine incorporation) as well as from quantitative histopathology in the liver of rats and mice. Two different approaches were used for hazard evaluation: (1) a "Safety Factor" approach based on no observed effect levels for liver tumors. and (2) calculation of lower confidence limits on risk-specific doses with the GLOBAL83 computer program. In each case, cytotoxicity produced by reactive CHCl3 metabolites was used as the measure of "dose" to the liver. The Safety Factor approach suggested that continuous exposure of human populations to concentrations of CHCl3 less than 2840 ppb in air or 13,900 ppb in water would not be likely to significantly increase the risk of developing liver tumors. The second approach suggested a "plausible upper 95% confidence limit" of 1 x 10(-5) for lifetime excess cancer risk for human populations continuously exposed to 2200 or 13,100 ppb CHCl3 in air or water, respectively.