The association between alcohol consumption and pancreatic cancer is unsettled. Altogether 243,169 men and women 20–79 years, without cancer at baseline, were followed with respect to pancreatic cancer by linkage to the Cancer Registry of Norway and the Norwegian Cause of Death Registry. They participated in a cardiovascular survey where information on alcohol consumption, smoking habits, anthropometric measures, and some biological variables were recorded. During 20 years of follow-up, 991 incident pancreatic cancers were registered. We estimated the hazard ratios with the Cox proportional hazards model, and graphed spline curves between glass-units/d of alcohol and hazard ratio of incident pancreatic cancer. The multivariable adjusted hazard per 1 glass-unit/d was 1.08 (95% confidence interval 1.02–1.15) for men and 1.04 (0.97–1.13) for women. The association between alcohol consumption and incident pancreatic cancer was present in ex- and current smokers, but the association could be ascribed to smoking habits. The multivariable adjusted spline curves increased with increasing glass-units/d and with confidence bands not encompassing 1.0 above one glass-unit/day. Our findings of an association between higher level of alcohol consumption and incident pancreatic cancer, could be attributed to confounding by smoking habits.