The epidemiological evidence linking Helicobacter pylori infection to stomach cancer is reviewed here in ascending order of epidemiological strength. Only those studies in which exposure to H. pylori was assessed by serology are considered. Ecological studies or geographical correlation studies have revealed inconsistent results when the incidence or mortality rates of stomach cancer have been correlated with the prevalence rates of H. pylori infection. Only those studies with a wide range in gastric cancer rates and in which a large number of populations are compared show weak but significant positive correlations. However, it is well known that these studies are difficult to interpret and do not provide strong evidence for a causal association. In only those case-control studies carried out in low-risk countries for gastric cancer, an increased risk for stomach cancer has been found associated with the presence of H. pylori antibodies. In high-risk countries for gastric cancer the lack of association may be explained by the use of inappropriate serological assays. Three case-control studies nested within cohorts have consistently shown an increased risk among subjects in whom the presence of H. pylori antibodies have been detected many years before the cancer diagnosis. They constitute the strongest evidence for a causal association. The final proof of causality will be provided by intervention studies in which it is demonstrated that long-term eradication of H. pylori is followed by a reduction in the gastric cancer rates. However, our own experience has shown that such studies are not easy to implement.