We studied the incidence of pancreatic cancer in 122,894 men and women who had previously reported amount and frequency of coffee, tea, and alcohol consumption; reporting was done at a multi-phasic health check-up (MHC) taken while subjects were members of a large prepaid health plan. We also tested the hypothesis that a pre-clinical effect of pancreatic cancer on glucose homeostasis leads to mild hyperglycemia and a generally increased thirst. If true, this could partially explain the increased consumption of beverages (particularly coffee) reported in association with pancreatic cancer in some case-control studies. However, in the 49 pancreatic cancer cases diagnosed during 6 years of follow-up, we found no evidence of increased risk associated with coffee, tea, or alcoholic beverages. We also found no evidence to support the increased-thirst hypothesis when we examined the 19 cases diagnosed within 12 months of having MHC. We did confirm a significantly increased risk among cigarette smokers (relative risk, 2.5; 95% confidence interval, 1.3-4.7) which was progressive with increasing levels of cigarette use. In addition, risk of pancreatic cancer was greater for persons previously under treatment for diabetes mellitus (relative risk, 4.5; 95% confidence interval, 1.2-16.7). Our results add to the growing body of evidence against a causal role of coffee in pancreatic cancer.