The authors report the results of a case-cohort study of 338 lung cancer deaths in 1950-1988 and a random sample (sub-cohort) of 1,138 from among 16,297 men who had worked at least one year between 1950 and 1979 in manual jobs at a large aluminum production plant. In the past, certain workers were exposed to substantial quantities of coal tar pitch volatiles, a mixture known to include polynuclear (polycyclic) aromatic hydrocarbons, and thus suspected to be capable of causing lung cancer. After they controlled for the effects of smoking, the authors found that rate ratios rose with cumulative exposure to coal tar pitch volatiles measured as benzene-soluble material to 2.25 (95% confidence interval (Cl) 1.50–3.38) at 10–19 mg/m3-years benzene-soluble matter, but did not rise further at higher exposures. The data are compatible with a linear relation with benzene-soluble matter (rate ratio (RR) = 1 + 0.031 mg/m3-years benzenesoluble matter). This model predicts a rate ratio of 1.25, and lifelong excess risk of 2.2%, after 40 years exposure at the current hygiene standard (0.2 mg/m3). A curved relation (RR = 1 + 0.098 mg/m3-years benzene-soluble matter 0.7) fitted somewhat better. Under this model, the predicted risks after this exposure are higher: 1.42 and 3.8%. The data are compatible with both additive and multiplicative models for the combined effect of smoking and coal tar pitch volatiles.