Data from a population-based, case-control study of 149 cases of cancer of the exocrine pancreas (excluding insulinomas) and 363 controls were used to test the hypothesis that coffee consumption increased the risk of cancer of the exocrine pancreas. Because of the high mortality from cancer of the pancreas, all information was obtained from proxy respondents for cases and controls. Increased risk was present for coffee drinkers [odds ratio (OR) = 2.38], cigarette smokers (OR = 2.27), and alcohol users (OR = 1.60), but the effect for alcohol disappeared after adjustment for cigarette use. Risk increased with the amount of coffee drunk with an OR of 2.38 in the those having at least 50,000 lifetime cups. This gradient was present in both men and women, and in cigarette smokers and nonsmokers. Increased risk was present for users of regular and decaffeinated coffee, but the risk was higher for users of decaffeinated coffee than users of regular coffee. Cigarette smoking, after controlling for coffee consumption, was an independent risk factor for pancreatic cancer, with odds ratios in the heaviest smokers of 2.71 (95% confidence interval = 1.41-5.21.