Results are reported with respect to smoking data from a population-based case-control study of pancreatic cancer conducted in Toronto, Canada, between 1983 and 1986. Lifetime smoking histories were obtained for 249 cases and 505 controls. A statistically highly significant positive association was observed between lifetime cigarette consumption and risk of pancreatic cancer. There was a rapid decrease in risk with time for those who quit cigarette smoking, the risk for ex-smokers being the same as for lifetime non-smokers between 10 and 15 years after quitting. Limiting exposure to the 15 years prior to diagnosis considerably strengthens the association and leads to a much more clearly defined dose-response relationship with relative risks of 1.88, 4.61, and 6.52 for tertiles of consumption for current cigarette smokers compared with lifetime non-smokers (p trend less than 10(-5)). We conclude that the current data, together with those from previous studies, strongly support a causal relationship between cigarette smoking and risk of pancreatic cancer, and indicate that cessation of smoking is likely to prove a rapidly effective preventive measure for this major type of cancer.