AIM: To validate a non-invasive method to detect gastric mucosal atrophy in a Chilean population with high prevalence of gastric cancer and a poor survival rate.METHODS: We first determined the optimal cut-off level of serum pepsinogen (PG)-1, PG-1/PG-2 ratio and 17-gastrin in 31 voluntary symptomatic patients (mean age: 66.1 years), of them 61% had histologically confirmed gastric atrophy. Then, in a population-based sample of 536 healthy individuals (209 residents in counties with higher relative risk and 327 residents in counties with lower relative risk for gastric cancer),we measured serum anti-Hpylori antibodies, PG and 17-gastrin and estimated their risk of gastric cancer.RESULTS: We found that serum PG-1 ＜ 61.5 μg/L,PG-1/PG-2 ratio ＜ 2.2 and 17-gastrin ＞ 13.3 pmol/L had a high specificity (91%-100%) and a fair sensitivity (56%-78%) to detect corpus-predominant atrophy.Based on low serum PG-1 and PG-1/PG-2 ratio together as diagnostic criteria, 12.5% of the asymptomatic subjects had corpus-predominant atrophy (0% of those under 25 years and 20.2% over 65 years old). The frequency of gastric atrophy was similar (12% vs 13%)but H pylori infection rate was slightly higher (77% vs 71%) in the high-risk compared to the low-risk counties.Based on their estimated gastric cancer risk, individuals were classified as: low-risk group (no H pylori infection and no atrophy; n = 115; 21.4%); moderate-risk group (H pylori infection but no atrophy; n = 354, 66.0%);and high-risk group (gastric atrophy, with or without H pylori infection; n = 67, 12.5%). The high-risk group was significantly older (mean age: 61.9 ± 13.3 years),more frequently men and less educated as compared with the low-risk group.CONCLUSION: We propose to concentrate on an upper gastrointestinal endoscopy for detection of early gastric cancer in the high-risk group. This intervention model could improve the poor prognosis of gastric cancer in Chile.