Objective To reasonably provide data basis of the expense standard for diagnosis related groups(DRGs)of respiratory system disease.Methods Use the first-page data of the medical records of 19,285 respiratory inpatients of a hospital.Use the Chi-squared Automatic Interaction Detector in the decision tree model to establish the inpatient case mix plan of respiratory disease inpatients.Then evaluate the model effectiveness with the variance reduction Results Expense median of the respiratory disease inpatients is 1603.97 yuan,averaging 2806.03 yuan,and the discretion degree is high.Proportion of surgical operations in respiratory disease is 3.89％,less than the others.Age is an important factor of respiratory disease for their hospitalization costs.Classification nodes of respiratory diseases are in turn the first discharge diagnosis,age,and secondary diagnosis.Patients were divided into 5 categories and 29 disease diagnosis related groups by such nodes.Conclusion Appropriate discharge diagnosis is the basis for building diagnosis related groups.Such variables as disease severity related variables as age and secondary diagnosis should serve as classification variables for respiratory system diseases.