Although declining, gastric cancer (GC) is estimated to be second in frequency worldwide. Major causes appear to be environmental rather than genetic. A relationship has been suggested between tobacco smoking and GC. A number of epidemiological studies have been performed dealing with this question. All the cohort studies showed a significantly increased risk of GC of the order of 1.5–2.5 for cigarette smokers. Evidence from case-control studies is less consistent. We have carried out a meta-analysis on the 40 studies providing a quantitative estimate of the association between GC risk and tobacco smoking. Results suggest a risk of stomach cancer among smokers of the order of 1.5–1.6 as compared to non-smokers. The summary relative risk was higher in men (1.59) than in women (1.11). Several studies examined the dose-response relationship which existed in 4 cohort studies and 6 case-control studies. We estimated the number of GC cases attributable to tobacco smoking occurring worldwide: in total, over 80,000 cases of GC (11% of all estimated cases) may be attributed to tobacco smoking each year. This figure is larger than that estimated for other cancers for which association with tobacco smoking is clearly established, such as pancreatic and renal cancers. Int. J. Cancer 72:565–573, 1997. © 1997 Wiley-Liss, Inc.