Chloroform increases the incidence of liver tumors in B6C3F1 mice when administered in by gavage in corn oil, but not when given in the drinking water at similar daily doses. Since cytotoxicity and regenerative cell proliferation have been implicated in the tumorigenic process for this nongenotoxic agent, these effects of chloroform in corn oil and drinking water were evaluated under conditions similar to the two bioassays. Female B6C3F1 mice were administered oral doses of 0, 3, 10, 34, 90, 238, or 477 mg/kg chloroform dissolved in corn oil 5 days/week for periods of 4 days or 3 weeks, or were continually exposed to chloroform in the drinking water at concentrations of 0, 60, 200, 400, 900, or 1800 ppm for 4 days or 3 weeks, at which time they were necropsied. 5-Bromo-2'-deoxyuridine (BrdU) was delivered via osmotic pumps implanted 3.5 days prior to necropsy. Cell proliferation was evaluated as the percentage of hepatocytes that entered S-phase over 3.5 days (labeling index, LI), measured by immunohistochemical detection of BrdU incorporated into the DNA. Dose-dependent changes included centrilobular necrosis and markedly elevated LI in mice given 238 or 477 mg/kg chloroform in corn oil (the average daily doses that produced tumors in the cancer bioassay). The no-observed-effect level for histopathological changes was 10 mg/kg/day and for induced cell proliferation was 34 mg/kg/day for chloroform given in corn oil. Chloroform given in the drinking water did not increase the hepatic LI after either 4 days or 3 weeks in any of the dose groups, nor were any microscopic alterations observed in the livers, even though the cumulative daily amount of chloroform ingested in the 1800-ppm exposure group was 329 mg/kg/day. The sustained increase in LI in the livers of mice administered hepatocarcinogenic doses of chloroform in corn oil, but not for chloroform in drinking water, is evidence that chloroform-induced mouse liver cancer is secondary to events associated with induced cytolethality and cell proliferation. The triggering of these effects appears to be dependent on both the rate and duration of chloroform delivery to the target tissues. Thus, the most straightforward risk assessment for chloroform for this tissue would assign no increased cancer risk for dosing regimens that do not induce cytolethality and cell proliferation.