Cancer risk assessments for trichloroethylene (TCE) based on linear extrapolation from bioassay results are questionable in light of new data on TCE's likely mechanism of action involving induced cytotoxicity, for which a threshold-type dose-response model may be more appropriate. Previous studies have shown that if a genotoxic mechanism for TCE is assumed, algebraic methods can considerably simplify the use of physiologically based pharmacokinetic (PBPK) models to estimate virtually safe environmental concentrations for humans based on rodent cancer-bioassay data. We show here how such methods can be extended to the case in which TCE is assumed to induce cancer via cytotoxicity, to estimate environmentally safe concentrations based on rodent toxicity data. These methods can be substituted for the numerical methods typically used to calculate PBPK-effective doses when these are defined as peak concentrations. We selected liver and kidney as plausible target tissues, based on an analysis of rodent TCE-bioassay data and on a review of related data bearing on mechanism. Tumor patterns in rodent bioassays are shown to be consistent with our estimates of PBPK-based, effective cytotoxic doses to mice and rats used in these studies. When used with a margin of exposure of 1000, our method yielded maximum concentration levels for TCE of 16 ppb (87 micrograms/m3) for TCE in air respired 24 hr/day, 700 ppb (3.8 mg/m3) for TCE in air respired for relatively brief daily periods (e.g., 0.5 hr while showering/bathing), and 210 micrograms/liter for TCE in drinking water assuming a daily 2-liter ingestion. Cytotoxic effective doses were also estimated for occupational respiratory exposures. These estimates indicate that the current OSHA permissible exposure limit for TCE would produce metabolite concentrations that exceed an acute no observed adverse effect level for hepatotoxicity in mice. On this basis, the OSHA TCE limit is not expected to be protective.