Increased body mass index has emerged as a potential risk factor for pancreatic cancer. The authors examined whether the association between body mass index and pancreatic cancer was modified by gender, smoking, and diabetes in residents of southeastern Michigan, 1996-1999. A total of 231 patients with newly diagnosed adenocarcinoma of the exocrine pancreas were compared with 388 general population controls. In-person interviews were conducted to ascertain information on demographic and lifestyle factors. Unconditional logistic regression models estimated the association between body mass index and pancreatic cancer. Males' risk for pancreatic cancer significantly increased with increasing body mass index (p(trend) = 0.048), while no relation was found for women (p(trend) = 0.37). Among nonsmokers, those in the highest category of body mass index were 3.3 times (95% confidence interval: 1.2, 9.2) more likely to have pancreatic cancer compared with those with low body mass index. In contrast, no relation was found for smokers (p(trend) = 0.94). While body mass index was not associated with pancreatic cancer risk among insulin users (p(trend) = 0.11), a significant increase in risk was seen in non-insulin users (p(trend) = 0.039). This well-designed, population-based study offered further evidence that increased body mass index is related to pancreatic cancer risk, especially for men and nonsmokers. In addition, body mass index may play a role in the etiology of pancreatic cancer even in the absence of diabetes.