The associations between smoking, caffeine, and alcohol intake and the risk of duodenal ulcer have rarely been investigated prospectively. We examined these associations in a prospective cohort of 47,806 men, 40-75 years of age, using a mailed baseline questionnaire in 1986, with follow-up every 2 years through 1992. During 6 years of follow-up, we documented 138 newly diagnosed cases of duodenal ulcer. After adjustment for age, energy-adjusted dietary fiber, body mass index, and use of aspirin or other nonsteroidal antiinflammatory drugs, current smoking was not associated with a substantial risk of duodenal ulcer [relative risk (RR) = 1.07; 95% confidence interval (CI) = 0.61-1.89]. Overall, past smokers were not at increased risk compared with never-smokers (RR = 0.99; 95% CI = 0.69-1.42). Adjusting for other risk factors, alcohol intake (comparing those who drink > 30 gm of alcohol per day to nondrinkers) was not associated with higher risk of duodenal ulcer (RR = 0.74; 95% CI = 0.42-1.29). We observed little association between caffeine, caffeine-containing beverages, and decaffeinated coffee and the risk of duodenal ulcer. These results indicate that smoking is not associated with a substantial increase in risk of duodenal ulcer, nor is high intake of alcohol and caffeine.