Objective To analyze the current situation and existing problems of the control type of randomized controlled trials （RCTs） and provide references and suggestions for improving the efficacy evaluation of RCTs. Methods The RCTs on Chinese herbal medicine for type 2 diabetes mellitus in China National Knowledge Internet,VIP database,Wanfang database,Sinomed,Pubmed and CENTRAL were searched from establishment of database to June 2017, and classified and evaluated properly according to control setting principle of clinical epidemiology. Results A total of 2046 RCTs published from 1986 to 2017 were included,including 2030 in Chinese and 16 in English. In this study, only 68 （3. 32 %） RCTs used placebo control. Glucoselowering western medicine and insulin were the main controls used by RCTs （1700,83. 09 %）, and the proportion increased with years. The proportion of Chinese herbal medicine as control was the second （207,10. 12 %）,but decreased during the last10 years. Conclusion Due to lack of high quality evidence in most of the Chinese herbal medicine at present, it may be unreasonable for Chinese herbal medicine as a control to prove the effect of interventions,and placebo-controlled trials should be added based on the conventional treatments.