To identify potential associations between workplace exposures and cancer mortality risks, job titles collected from 1965 to 1971 for 58,678 men (a subset of a large representative sample of the Canadian workforce) were transformed into probable chemical exposures using a job-exposure matrix developed in Montreal. Mortality follow-up was determined through computerized record linkage with the National Mortality Database in Canada for 1965-1991. Cancer mortality risk was evaluated at two levels of exposure, any and substantial, using Poisson regression controlling for age, calendar period, and social class. Among the 58,678 men, 3,160 died of cancer. Using a liberal reporting criterion, relative risk (RR) >1.0, five or more exposed cancer deaths, p < or = 0.100, several potential associations were identified, including: lung cancer and any exposure to abrasives dust (RR = 2.84), prostate cancer and any exposure to calcium carbonate (RR = 2.46), and prostate cancer and substantial exposure to metallic dust (RR = 2.13).