Abstract: Objective To Understand the quality of graduation thesis of medical graduate students, and find ways to improve the quality of graduate thesis through systematic training of clinical epidemiology and evidence-based medical knowledge. Methods Taking the graduate students of a medical school in Zhengzhou from 2014 to 2017 as the survey object, using the STARD (2003) report specification of the diagnostic test to evaluate the clinical epidemiological and evidence-based medical knowledge system before and after training, the graduate diagnostic test paper STARD (2003) reported 25 items in line with the situation, and judged the effect of clinical epidemiology. Results A total of 77 postgraduate diagnostic test papers that met the inclusion criteria, of which 23 were not trained in clinical epidemiological methods, 54 were trained in clinical epidemiological methods, and received clinical epidemiology and follow-up. After the training of the relevant knowledge system of the medical science, the coincidence rate of the STARD report items of the graduate diagnostic test papers was significantly improved, and the average number of entries increased from 9.36±2.25 to 18.64±2.46. The difference was statistically significant (t=15.527, P<0.001), and the entries rate was higher than the pre-training study, and the 80% entry rate difference was statistically significant (P<0.05). The compliance rate of the 15th item increased by 72.7%. Conclusion Systematic training of clinical epidemiology and evidence-based medical knowledge can significantly improve the quality of graduate thesis and is worthy of further application and promotion.