Background: Absence of national US gastric cancer (GC) screening necessitates alternative methods like surveys to identify high-risk populations. Identification of high-risk persons may be enhanced by adding ethnic and cultural variables to more conventionally known risk factors for GC. Methods: Data from a prior case-control study of 40 GC cases and 100 controls were used. A "conventional" risk factor model (age, gender, family history of GC, body mass index, excessive salt intake, alcohol, smoking, blood type, H pylori) was compared to one incorporating ethnic and cultural variables (race, immigration, generation, cultural food at ages 15-18 years, acculturation and education) using model fit, sensitivity, specificity and expected positive predictive values (PPV). Stepwise regression was then used to create a model from this pool of variables. PPV was calculated using Bayes' Theorem applied to the baseline GC incidence in the US (7.2 per 100,000). Results: The "conventional" model required 14 questions and resulted in 25% sensitivity, 94% specificity, 28 per 100,000 PPV at the 70% probability cut-off, and AUC=0.871. The model incorporating ethnic and cultural variables required 38 questions and resulted in 48% sensitivity, 91% specificity, 38 per 100,000 PPV, and AUC =0.965. After eliminating items less predictive at p=0.2, age, gender, family history of GC, excessive salt intake, immigration, generation and race remained in the model. This model required 7 questions and resulted in 45% sensitivity, 96% specificity, 81 per 100,000 PPV and AUC=0.914. Conclusions: The model with the greatest ability to identify persons at risk of GC included ethnic and cultural variables. This model can be translated into a survey with few items that can serve as a highly scalable tool to identify high-risk individuals. Support: UG1CA189823