Objective To investigate the efficacy and safety of metformin combined with insulin for treatment of elderly patients with type 2 diabetes mellitus (T2DM). Methods A total of 52 elderly T2DM patients (38 males and 14 females, aged from 75 to 89 years) hospitalized in our department from March 2016 to August 2017 were enrolled in this study. They were randomly divided into the metformin group and control group (n=26 for each group). Insulin detemir and insulin aspart were given to the control group, and the insulin dosage was adjusted according to the blood glucose level. In the metformin group, metformin was added to insulin therapy. The treatment continued for 12 weeks. Blood glucose control, blood glucose fluctuation, liver and kidney dysfunction were observed before and after treatment in order to evaluate the therapeutic effect and safety. SPSS statistics 19.0 was used to perform the statistical analysis. Student’s t test, Chi-square test, or Wilcoxon rank sum test was employed for comparison between groups. Results Compared to before treatment, the patients in the metformin group had their FBG [7.33(6.21,9.25) vs 8.33(6.23,0.89)mmol/L] and postprandial 2h blood glucose [8.35(7.31,9.80) vs 9.48(8.28,1.40)mmol/L] significantly decreased, so were those in the control group [7.60(6.10,0.10) vs 8.60(6.80,1.80)mmol/L, 10.32(7.31,1.20) vs 11.62(8.37,3.39)mmol/L] after 12 weeks of treatment (all P<0.05). In control group, there were no obvious differences in insulin dosage of 24 h, HbA1c, standard deviation of blood glucose (SDBG), mean amplitude of glycemic excursion (MAGE), largest amplitude of glycemic excursion (LAGE), and mean postprandial plasma of glycemic excursions (MPPGE) after treatment (P＞0.05). While, in the metformin group, HbA1c, SDBG, MAGE, LAGE, MPPGE, and insulin dosage of 24 h were remarkable decreased, and these indices were notably lower than those of the control group (P<0.05). During the treatment, hypoglycemic events occurred for 7 times in 6 cases. They all got remission after eating some food. No serious hypoglycemic event was observed. The incidence rate of hypoglycemia was significantly lower in the metformin group than the control group [7.69%(2/26) vs 19.23%(5/26), χ²=6.892, P<0.05]. Conclusion Metformin can be used in the elderly T2DM patients in case of blood glucose poorly controlled by insulin treatment, and it can control blood glucose effectively and stably, with high safety.