The power of an SNP-based genome-wide association study (GWAS) was first demonstrated in Japan using the JSNP database and is currently a major strategy adopted around the world for a number of common diseases including cancers. The hypothesis-free strategy can lead us to a novel hypothesis for carcinogenesis and may contribute to identifying a high risk group for research and, in the future, practice of personalized prevention. We performed a GWAS on diffuse-type gastric cancer and identified a significant association with SNPs in the PSCA (prostate stem cell antigen) gene. The association was validated by a Korean gastric case-control analysis. The PSCA protein is expressed predominantly in the stem cell/precursor-rich region of the gastric epithelium, which is considered as the origin of diffuse-type gastric cancer, and showed tumor suppressor-like characteristics. Individuals with a low PSCA promoter activity are susceptible to diffuse-type gastric cancer. By contrast, the polymorphism does not significantly predispose to intestinal-type gastric cancer, congruous to the hypothesis of the two distinct carcinogenesis pathways for the two major types of gastric cancer. In addition to publication on a specific gene, the sharing of GWAS data through a database on the web is expected to accelerate validation and discovery by other investigators. GeMDBJ (Genome Medicine Database of Japan), started in 2005 in Japan, is one of such attempts. Moreover, the advent of “next generation” sequencers may herald a new era in which the poorly explored domains of the genetic architecture of disease susceptibility may be unveiled. (Cancer Sci 2010)