An analysis of the mortality experience of workers exposed to dinitrotoluene (DNT) was conducted to test the hypothesis that DNT exposure is associated with an increased risk of cancers of the liver and biliary tract. A total of 4,989 workers exposed to DNT and 7,436 unexposed workers who had worked for at least 5 months at the study facility between January 1, 1949 and January 21, 1980, were included in this investigation. Workers were considered exposed if they had worked at least 1 day on a job with probable exposure to DNT. The vital status as of December 31, 1982, was successfully ascertained for approximately 97% of these workers. Standardized mortality ratios (SMRs) were estimated based upon comparisons with the US population using a modified life-table program. In addition, standardized rate ratios (SRRs) were computed based upon direct comparisons between the DNT and the internal unexposed cohort. An excess of hepatobiliary cancer was observed among workers exposed to DNT in this study. The rate ratio for hepatobiliary cancer was 2.67 (six cases observed) based upon comparison with the US population (SMR = 2.67, 95% CI = 0.98, 5.83), and 3.88 based upon comparison using the internal unexposed referent group (SRR = 3.88, 95% CI = 1.04, 14.41). This study failed to demonstrate an exposure-response relationship between duration of DNT exposure and hepatobiliary cancer mortality. Our study was limited by the small number of workers with long duration of exposure to DNT, and by the lack of quantitative information on exposure to DNT and other chemicals.(ABSTRACT TRUNCATED AT 250 WORDS)