A diagnosis model for type 2 diabetes mellitus based on laboratory test results was developed by the relationships among the principal parameters. Sample data were collected from a specialized diabetes mellitus clinic located in Korea. The total number of subjects was 953. Statistical analyses including the t-test were used to select the principal parameters, and a decision tree and clustering methods including expectation maximization were used to investigate the relationships among the principal parameters. This study obtained ten statistically principal parameters among 32 laboratory test parameters and suggested four different groups, the patient group, high-probability group, low-probability group, and normal group. Although fasting glucose is meaningful to the diagnosis of diabetes mellitus, six critical parameters are especially important to ensure a more reliable diagnosis. This study will provide a useful resource for diabetes research in Korea.