This study was carried out to develop a scoring system for the diagnosis of gastric adenocarcinoma (GAC). A total of 686 subjects, 150 patients with GAC, 182 with gastric ulcer, 127 with duodenal ulcer, and 227 subjects with negative findings, were enrolled. Analysis of the likelihood ratio (LR) showed that patients with advanced age, ulcer in the stomach, low serum levels of pepsinogen I (PGI), low PGI x gastrin values, and low PGI/gastrin ratio were likely to have GAC. Of these indicators, the serum PGI level had the greatest weight, with a LR of 7.59 for the group with a level < 30 ng/ml. A scoring system combining serum PGI level, Helicobacter pylori seropositivity, and gastric ulcer status was derived, using a logistic regression model. This scoring system was found to be better than any one-parameter criterion for diagnosing GAC after evaluation by the area under the receiver operating characteristic curve (0.84; 95% confidence interval, 0.81-0.88) or by specificity-fixed sensitivity (sensitivity 0.82 at specificity 0.72, sensitivity 0.87 at specificity 0.66, sensitivity 0.96 at specificity 0.44). This scoring system may be potentially useful as a new model for the noninvasive diagnosis of GAC in the future.