Abstract: Objective： The purpose of this paper is to compare the effectiveness of metformin with acarbose and metformin with repaglinide under non-randomized trails. Methods： Medical records of inpatients in endocrinology department in a Beijing′s hospital were analyzed， we collect the demographic data and conditions of change of index， and patients were divided into metformin with acarbose group（group A）and metformin with repaglinide group （group B）. We use frequencies analysis method to deal enumeration data， as well as chi-square test and Wilcoxon rank sum test to ensure the difference；We use mean ± standard deviation （±s）to deal measurement data， as well as paired t-test and Wilcoxon signed ranks test to ensure difference.Results：In the comparison of effectiveness in the same group， Patients in group A suggested that there was no statistically significant difference in two hour postprandial blood glucose（2h PBG）before and after treatment （P>0.05）； fasting blood-glucose（FBG）（P<0.05）and glycosylated hemoglobin （HbA1c）（P<0.05）had a great improvement. At the same time， the 2h PBG， FBG and HbA1c had a great improvement in group B（P<0.05）. In the comparison of effectiveness in the different group， the improvement of HbA1c in group A was better than group B. Conclusion：The historic cohort studies have a high practical reference value； metformin with repaglinide is better than metformin with acarbose in cure process； insulin secretagogue drugs and glucose absorption inhibitor drugs may have better effectiveness than other plan； the confounding factors may reduce the level of effectiveness differences.