The etiology of pancreatic cancer is poorly understood, partly because of the inconsistency of findings among case-control studies of pancreatic cancer. Because of the unfavorable prognosis for pancreatic cancer, many case-control studies have been based largely on interviews with next of kin, who are known to report less reliable information on potential risk factors than original respondents. The purpose of this study was to estimate the effects of speculative risk factors such as dietary/nutritional factors and alcohol drinking, as well as those of established risk factors such as cigarette smoking, diabetes mellitus, and family history of pancreatic cancer, on pancreatic cancer risk based solely on direct interviews. This investigation was a population-based case-control study of pancreatic cancer diagnosed in Atlanta (GA), Detroit (MI), and ten New Jersey counties from August 1986 through April 1989. Direct interviews were conducted with 526 incident cases and 2,153 population controls. This study revealed a significant interaction between body mass index and caloric intake that was consistent by both race and gender. Subjects with elevated body mass index and caloric intake had increased risk, whereas those with elevated values for one of these factors but not the other experienced no increased risk. This finding suggests that energy balance may play a major role in pancreatic carcinogenesis. Diabetes mellitus was also a risk factor for pancreatic cancer, as well as a possible complication of the tumor. Our data are consistent with a key role for hyperinsulinemia in pancreatic carcinogenesis, particularly among non-diabetics with an elevated body mass index. A three-fold risk of pancreatic cancer among first-degree relatives of affected individuals was apparent. An increased risk also was associated with a family history of colon, endometrial, ovary, and breast cancer, suggesting a possible link to hereditary non-polyposis colon cancer. Our findings support a causal role for cigarette smoking in pancreatic carcinogenesis. Alcohol drinking at levels typically consumed by the general population of the United States did not appear to be a risk factor for pancreatic cancer, although heavy drinking may be related to risk, particularly in blacks.