The relation between various aspects of medical history, selected indicator foods and the risk of pancreatic cancer was analyzed in a hospital-based case-control study conducted in Northern Italy on 247 patients with cancer of the pancreas, and 1,089 controls in hospitals for acute, nonneoplastic or digestive conditions. There was a significant association with history of pancreatitis (relative risk, RR 3.2, 95% confidence interval = 1.3-7.9), which was however reduced when the condition was first diagnosed at least 5 years previously. The point estimates were slightly, but not significantly, above unity for diabetes (RR = 1.5), gastrectomy (RR = 1.1) and cholelithiasis (RR = 1.3), and no association was found with liver disease or drug allergy. In relation to diet, there was some tendency for the risk to decrease with more frequent fruit consumption, but the results were largely inconsistent in relation to various indicators of meat, animal protein or fat intake. Although no important associations were found in this study with various aspects of medical history or diet indicators and pancreatic cancer risk, on account of the size of the dataset and the statistical power, this study contributes usefully to the debate on a common cancer whose causes are still largely undefined.