Microsatellite instability (MSI) defines a specific type of genetic instability. Although consensus diagnostic criteria for MSI definition in colorectal cancer have been established, their utility in other tumor types remain to be proven. Previously we developed a mathematical model for MSI definition in colorectal cancer. The aim of this study was to establish diagnostic criteria for MSI evaluation in human gastric cancer. We designed an algorithm for the efficient characterization of MSI and used it to analyze data on 7 microsatellite markers in 35 gastric carcinomas. Theoretical models considering 1, 2, or 3 populations were tested against the data collected. Also, hypermethylation of hMLH1 gene promoter and hMLH1 protein expression were studied. The observed frequencies of MSI in our series of samples best fit a 2-population model: stable and unstable, defined by instability in 2 or more of a minimum of 7 markers analyzed. MSI was observed in 29% of the tumors. Misclassification rate was <4% when any 7 loci were analyzed. MSI(+) tumors inversely associated with p53 protein overexpression. A good correlation between hMLH1 status (either protein or promoter hypermethylation) and MSI classification was observed. We have developed a simple, sensitive, and specific approach to assess the presence of MSI in gastric cancer that may have clinical applications.