To offer quantitative evidence on the association between pancreatitis and pancreatic cancer, we analyzed data from a hospital-based case-control study conducted in northern Italy between 1983 and 1992, including a total of 362 incident cases of histologically confirmed pancreatic cancer and 1,408 controls admitted to hospital for acute, nonneoplastic, nondigestive tract disorders. Information was obtained using a structured questionnaire on sociodemographic characteristics and lifestyle habits (including tobacco and alcohol consumption) and a problem-oriented medical history, which included history of pancreatitis and age at its first diagnosis. Pancreatitis was reported by 24 (6.6%) cases and 18 (1.3%) controls, yielding an age- and sex-adjusted relative risk (RR) of 5.7 (95% confidence interval, 2.9-11.4). The risk of pancreatic cancer was appreciably higher 5 or more years after diagnosis of pancreatitis (RR = 6.9) than in the first 4 years (RR = 2.1), and in subjects below age 60 (RR = 8.3) than in elderly ones (RR = 2.6), but similar in males and females. The time-risk relationship is strongly indicative of a real relationship between pancreatitis and pancreatic cancer. After allowing for tobacco and alcohol, besides area of residence and education, the association between pancreatitis and pancreatic cancer appeared to be partly explained by such covariates (RR = 3.9); however, this may represent some degree of overadjustment if, for instance, alcohol is causally linked to pancreatitis, which, in turn, is causally related to pancreatic cancer. In terms of population attributable risk, pancreatitis would explain approximately 5% of pancreatic cancer cases.