A hospital-based, incident case-control study of pancreatic cancer was conducted between 1979 and 1983 in parallel with similarly designed studies of lung and stomach cancers in high-risk areas of Louisiana. To evaluate life-style practices, including diet, the authors pooled controls from the three studies, and then excluded subjects with diet-altering chronic diseases. When the 363 cases were compared with the 1,234 identified controls, significantly elevated risks were found among persons with Cajun ancestry, especially in rural areas. Among current smokers, a significant twofold risk was associated with moderate (16-25 cigarettes per day) and heavy (greater than or equal to 26 cigarettes per day) consumption, while ex-smokers showed no consistent pattern of risk. After adjustment for potential confounding by smoking, diet, and demographic factors, the risk of pancreatic cancer was unrelated to use of alcoholic beverages or coffee. Pork products and rice were conspicuous as dietary risk factors, each showing a positive dose-response effect, whereas fruit consumption exerted a protective influence.