Objective: To explore the optimal grouping scheme of psychiatric Diagnosis Related Groups (DRG), and to provide sup- port for improving the targeting and effectiveness of DRG tools in psychiatric medical insurance payments. Methods: Based on the Beijing-Tianjin-Hebei Mental health big data platform, data of inpatients eligible for DRG inclusion in Beijing Anding Hospital from 2018 to 2019 were selected for retrospective study. An optimal grouping scheme was constructed based on the decision tree model. Variance analysis and rank sum test were used to compare the differences within and between MDCT-DRG groups, and reduction in- variance (RIV) was used to evaluate the grouping efficiency. Results: After the optimization, the intra-group consistency and in- ter-group heterogeneity of psychiatric DRG were improved, and the grouping efficacy was better. Conclusion: The DRG grouping scheme in Psychiatric Department can be optimized by decision tree model. In the national medical insurance reform, the characteris- tics of hospitalization and consumption of medical resources by psychiatric patients should be further studied, to formulate a more suit- able payment method for psychiatric departments, so as to play a more effectively role of DRG in medical insurance and performance evaluation.