We studied 50 lung cancer cases and 154 matched referents, all dock workers, for whom we obtained smoking information and employment histories. We assessed exposures from information on annual diesel fuel consumption from each of the 15 ports included. We used a smoker/nonsmoker term and three exposure variables (machine time, cumulative fuel, and exposed time with fuel consumption above a minimum cutpoint) in the analyses, with three categories for each exposure variable. Odds ratios (ORs) for medium and high exposure groups are consistently higher than reference (low), with an increasing exposure-response trend that is most marked for the exposed time variable (ORs: low = 1.0; medium = 1.6; high = 2.8). When smoking and that exposure variable are simultaneously included in the analyses, odds ratios for the medium (OR = 2.7) and high (OR = 6.8) levels of exposure increase, as does the odds ratio for smoking. Separating smokers and nonsmokers, with the low exposed nonsmokers as the common reference category, the odds ratios are 1.6 (medium) and 2.9 (high) for the nonsmokers, and 10.7 (medium) and 28.9 (high) for smokers. These results indicate an independent effect of diesel exhaust exposure and a strong interaction between smoking and diesel exhaust.