A cohort study of lung cancer was conducted among 7,961 Japanese-American men who were interviewed and examined during 1965-1968. Information was collected about their smoking history, occupation, and nutrient intake. After 22 y, 227 incident cases of lung cancer were identified. Cigarette smoking significantly increased lung cancer risk. The relative risk (RR) was 3.1 for past smokers and 11.4 for current smokers, compared with never smokers. We separated lung cancer cases according to histological type, and it was found that current smokers had a RRs of 16.0 for squamous/small-cell carcinoma and 6.8 for adenocarcinoma of the lung. Unskilled manual workers had a significantly higher risk (RR = 1.5; 95% confidence interval, 1.1-2.2) for lung cancer than workers who were engaged in nonmanual occupations. There was no association between lung cancer and the 24-h intake of total calories, protein, fat, dietary cholesterol, carbohydrates, and alcohol, but this may have resulted from the limitations of a 24-h dietary questionnaire.