OBJECTIVE：To analyze th e general characteristics and application of the models used in the pharmacoeconomic evaluation of type 2 diabetes mellitus （T2DM）therapy during the past 5 years，and to provide reference for the selection and improvement of T 2DM pharmacoeconomic model. METHODS ：Retrieved from PubMed ，Embase，CNKI，Wanfang database and VIP during Jan. 1st，2015 to Dec. 31st，2019，pharmacoeconomic evaluation literatures about T 2DM therapy were collected ；the included model was analyzed in respects of general structure ，therapy plan establishment ， short-term therapeutic efficacy ， complication simulation ，model effectiveness validation ，application frequency. RESULTS ：A total of 81 literatures were included ， involving 14 models，such as CORE model ，Cardiff model ，ECHO model ，etc. Mostly ，Markov or micro Markov simulation method were adopted to measure the patient ’s lifetime health outcome and cost mostly from the point of view of third-party payer. Seven models could simulate 2-4 therapy plans ；short-term efficacy mainly included risk factors of diabetic complications （such as glycosylated hemoglobin level and body mass ）and adverse drug reactions. Most models used intermediate indexes to simulate the occurrence of complications ，and the number of complications ranged from 3 to 15；the validity of model effectiveness included surface validity ，internal validity and external validity ，etc. Among 14 models，the most frequently used models in the past 5 years were long-term models that had been validated ，among which CORE model had the most application times （38/81，46.91％）， followed by Cardiff model （12/81，14.81％）. CONCLUSIONS ：The 14 models have similar structure. The differences of the models are mainly reflected in 3 aspects as therapy plan setting， considered short-term efficacy ， the number of model are ideally choose based on available evidences.