A multi-centre case-control study of pancreas cancer, designed to be population-based, to use a random sample of local populations as controls and to use a common protocol and core questionnaire, was conducted as the first study of the SEARCH programme of the International Agency for Research on Cancer. "Ever-smokers" were found to be at increased risk for pancreas cancer compared with "never-smokers" consistently in all strata of gender, response status and centre. Risk of pancreas cancer was found to increase with increasing lifetime consumption of cigarettes, the relative risk rising to 2.70 (95% C.I. 1.95 to 3.74) in the highest intake category. The overall trend in risk was highly significant and the association was found consistently in each stratum of gender, response status and centre. Fifteen years had to pass from quitting cigarette smoking until the risk fell to a level compatible with that in never-smokers among the heaviest group of smokers; among the 2 lowest tertiles this happened within 5 years. Further, reported smoking habits more than 15 years before diagnosis appeared to have no influence on pancreas-cancer risk, irrespective of amount smoked. The results are consistent with a causal role for cigarette smoking in the aetiology of pancreas cancer and illustrate that ceasing to smoke cigarettes can lead to reductions in the elevated risk of pancreas cancer produced by this habit.