OBJECTIVES: We studied relationships of cigarette smoking and coffee drinking to risk of pancreatitis. METHODS: This was a cohort study among 129,000 prepaid health plan members who supplied data about demographics and habits in 1978-85. Among 439 persons subsequently hospitalized for pancreatitis, probable etiologic associations were cholelithiasis (168/439 = 38%), alcohol (125/439 = 29%), idiopathic (110/430 = 25%), and miscellaneous (36/439 = 8%). Cox proportional hazards models with seven covariates (including alcohol intake) yielded relative risk estimates for smoking and coffee use. RESULTS: Increasing smoking was strongly related to increased risk of alcohol-associated pancreatitis, less related to idiopathic pancreatitis, and unrelated to gallstone-associated pancreatitis. Relative risks (95% confidence intervals, CI) of one pack per day (vs never) smokers for pancreatitis groups were: alcohol = 4.9 (2.2-11.2, p < 0.001), idiopathic = 3.1 (1.4-7.2, p < 0.01), and gallstone = 1.3 (0.6-3.1). The relationship of smoking to alcohol-associated pancreatitis was consistent in sex and race subsets. Drinking coffee, but not tea, was weakly inversely related to risk only of alcohol-associated pancreatitis, with relative risk (95% CI) per cup per day = 0.85 (0.77-0.95; p= 0.003). Male sex, black ethnicity, and lower-educational attainment were other predictors of alcohol-associated pancreatitis. CONCLUSIONS: Cigarette smoking is an independent risk factor for alcohol-associated and idiopathic pancreatitis. Coffee drinking is associated with reduced risk of alcohol-associated pancreatitis. The data are compatible with the hypotheses that smoking may be toxic to the pancreas or may potentiate other pancreatic toxins while some ingredient in coffee may have a modulating effect.