[Objective] To explore the early impact of diabetes mellitus (DM) on total knee arthroplasty (TKA) . [Methods] A retrospec- tive study was conducted on 252 patients who underwent primary TKA for knee osteoarthritis in our hospital. The patients were divided into two groups according to whether they had diabetes or not, including 138 cases in the non-DM group and 114 cases in the DM group. The ex- tents of knee swelling, pain and function were evaluated and compared between the two groups. [Results] Although there were no signifi- cant differences in operative time, incision length and postoperative drainage between the two groups (P＜0.05) , the non-DM group proved significantly superior to the DM group in terms of intraoperative blood loss, wound healing and hospital stay (P＜0.05) . All patients were fol- lowed up for 6~8 months with an average of (6.56±1.35) months. The VAS score and bilateral difference of knee circumference significantly decreased (P＜0.05) , whereas the knee ROM and HSS significantly increased in both groups over time (P＜0.05) . The non-DM group was significantly superior to the DM group in abovementioned items from 1 to 3 months postoperatively (P＜0.05) , despite of the fact that the dif- ferences between the two group in VAS, ROM and HSS score became not statistically significant at 6 months postoperatively (P＞0.05) . Ra- diographically, the femorotibial angle (FTA) and medial proximal tibial angle (MPTA) significantly improved (P＜0.05) , whereas the posteri- or tibial slope (PTS) remained unchanged postoperatively in both groups compared with those preoperatively (P＜0.05) . However, there were no statistically significant differences in FTA, MPTA and PTS between the two groups at any corresponding time points (P＞0.05) . [Conclusion] The DM does impair early postoperative functional recovery of TKA, with increasing affected knee swelling, reducing the range of mo- tion and functional score of affected knee, and increasing the risk of postoperative complications.