A population-based case-control study on lung cancer was conducted in 2 industrialized areas of northern Italy. Cases (126) were all males who died from lung cancer between 1976 and 1980. Controls (384) were a random sample of males dying from other causes during the same period. Jobs held during working life have been analyzed according to a list of occupations already known to be causally associated with lung cancer (list A) and a list of occupations suspected of being so (list B). Attributable risk percentages in the population for occupations included in either list A or B were about 36% and 12% in the 2 areas. Welders or workers in industries in which welding is common showed elevated odds ratios: 2.9 for welders (95% CI 0.9-9.8); 4.9 (1.1-22.9) for structural metal workers; 11.4 (2.6-49.9) for workers in structural metal production. Other job categories associated with lung cancer included: electricians and workers in electrical machine production, woodworkers (in furniture or cabinet making, but not in carpentry or joinery) and cleaning services. Smoking did not seem to exert a substantial confounding effect. Attributable risk percentages for tobacco smoking were about 78% and 76% in the population of the 2 areas.