In a case-control study, 839 male hospital-based cases of primary lung cancer and the same number of population-based controls--matched by sex, age, and region of residence--were personally interviewed for their job and smoking histories. The study allows to quantify occupational asbestos exposure that was thought to be a welding-associated risk: 6% of cases and 2% of controls were classified into the occupational category "welders or burners# (odds ratio [OR] = 2.65). This OR was reduced to 1.93 (95% confidence limit [CL]: 1.03-3.61) after adjustment for smoking and asbestos. In contrast, a history of welding in general for at least a half-year is 28% among cases and 23% among controls, yielding an OR of 1.25 (95% CL: 0.94-1.65) after adjustment for both confounders. The OR of welding for more than 6,000 hr is 1.45 (95% CL = 1.04-2.02), reduced to 1.10 after adjustment for smoking and asbestos. Oxyacetylene welding for more than 6,000 hr lifelong is associated with an OR of 1.86 (95% CL = 1.01-3.43) reduced to 1.46 (n.s.) after adjustment for smoking and asbestos. The risk of oxyacetylene welding seems to be highest for oat cell carcinoma with an adjusted OR for ever-exposure of 1.46 (95% CL = 0.69-3.10). Therefore, the present study supports the hypothesis that some, but not all, of the excess risk of welders observed in the literature may be due to a history of cigarette smoking and occupational asbestos exposure. The elevated risk for the subgroup of employees in the aircraft industry reported for the midterm evaluation of the study still prevails, though no longer statistically significant. However, employees in this industry who ever welded show an OR of 2.29 (95% CL = 1.19-4.42) after adjustment for smoking and asbestos.