Smoking has recently been recognised as causally associated with the development of gastric cancer (GC). However, evidence on the effect by sex, duration and intensity of smoking, anatomic subsite and cessation of smoking is limited. Our objective was to assess the relation between tobacco use and GC incidence in the European Prospective Investigation into Cancer and Nutrition (EPIC). We studied data from 521,468 individuals recruited from 10 European countries taking part in the EPIC study. Participants completed lifestyle questionnaires that included questions on lifetime consumption of tobacco and diet in 1991–1998. Participants were followed until September 2002, and during that period 305 cases of stomach cancer were identified. After exclusions, 274 were eligible for the analysis, using the Cox proportional hazard model. After adjustment for educational level, consumption of fresh fruit, vegetables and preserved meat, alcohol intake and body mass index (BMI), there was a significant association between cigarette smoking and gastric cancer risk: the hazard ratio (HR) for ever smokers was 1.45 (95% confidence interval [CI] = 1.08–1.94). The HR of current cigarette smoking was 1.73 (95% CI = 1.06–2.83) in males and 1.87 (95% CI = 1.12–3.12) in females. Hazard ratios increased with intensity and duration of cigarette smoked. A significant decrease of risk was observed after 10 years of quitting smoking. A preliminary analysis of 121 cases with identified anatomic site showed that current cigarette smokers had a higher HR of GC in the cardia (HR = 4.10) than in the distal part of the stomach (HR = 1.94). In this cohort, 17.6 % (95% CI = 10.5–29.5 %) of GC cases may be attributable to smoking. Findings from this large study support the causal relation between smoking and gastric cancer in this European population. Stomach cancer should be added to the burden of diseases caused by smoking. © 2003 Wiley-Liss, Inc.