The objective of the current study was to estimate the risk of lung cancer attributable to occupational factors and not due to tobacco. At 24 hospitals in nine metropolitan areas in the United States, 1793 male lung cancer cases were matched for race, age, hospital, year of interview, and cigarette smoking (never smoker, ex-smoker, smoker (1-19 and > or = 20 cigarettes per day)) to two types of controls (cancer and non-cancer hospital patients). Information on usual occupation, exposure to specific potential carcinogens, and cigarette smoking was obtained by interview. Risk of lung cancer was increased significantly for electricians; sheetmetal workers and tinsmiths; bookbinders and related printing trade workers; cranemen, derrickmen, and hoistmen; moulders, heat treaters, annealers and other heated metal workers; and construction labourers. All of these occupations are potentially exposed to known carcinogens. Odds ratios (ORs) were increased for exposure to coal dust (adjusted OR = 1.5; 95% confidence interval (95% CI) 1.1-2.1). After stratification, this association was statistically significant only after 10 or more years of exposure. Lung cancer was also related to exposure to asbestos (adjusted OR = 1.8; 95% CI 1.5-2.2). The ORs increased with increasing duration of exposure to asbestos for all smoking categories except for current smokers of 1-19 cigarettes per day. The statistical power to detect ORs among occupations that were previously reported to be at increased risk of lung cancer but that failed to show an OR of at least 1.5 in the current study was small. The cumulative population attributable risk (PAR) of lung cancer due to occupation was 9.2%. It is concluded that occupational factors play an important part in the development of lung cancer independently of cigarette smoking. Because occupations at high risk of lung cancer were under-represented, the cumulative PAR of the present study is likely to be an underestimate of the true contribution of occupation to risk of lung cancer.