The findings of a case - control study of cancer of the pancreas, which was conducted in the Baltimore metropolitan area, are reported. Two hundred one patients with pancreatic cancer were matched on age (+/- 5 years), race, and sex to hospital and non-hospital controls, the latter selected by random-digit-dialing (RDD). All subjects were interviewed regarding diet, beverage consumption, occupational and environmental exposures, and medical and surgical history. Significantly decreased risks were associated with consumption of raw fruits and vegetables and diet soda, and significantly increased risks were associated with consumption of white bread when cases were compared with hospital and RDD controls. A significantly reduced risk was associated with consumption of wine when cases were compared to RDD controls. Risk ratios for consumption of coffee were not significantly different from one, although there appeared to be a dose - response relationship in women. A moderate but statistically nonsignificant increase in relative odds was found for cigarette smoking, and cessation of smoking was associated with a marked reduction in risk. No significant associations were found with particular occupational exposures. Tonsillectomy was associated with a significantly reduced risk, a finding that has been observed for other cancers as well. The current evidence indicates that pancreatic cancer is likely to result from a complex interaction of factors and suggests that the study of its etiology requires a multidisciplinary approach involving both laboratory and epidemiologic components.