Results are reported from a population-based study of 249 cases of pancreas cancer and 505 controls carried out in Toronto, Canada, between 1983 and 1986. Lifetime consumption of coffee and alcohol and medical histories were assessed by personal interviews. No evidence of any association was found with different types of coffee or alcohol after adjusting for smoking, calories and fibre intake. There was a significant increased risk associated with a history of diabetes mellitus within 5 years of cancer development. A protective effect of a history of some allergic conditions, hay fever, eczema and asthma, was observed, although the relative risks were not significant (p value greater than 0.10).