To explore clinical efficacy and safety of application of tranexamic acid for two times combined with postoperative hip and knee on flexion position for reducing blood loss during total knee arthroplasty.From January 2015 to January 2018, 90 patients with knee osteoarthritis underwent unilateral total knee arthroplasty, including 33 males and 57 females, aged from 61 to 85 years old with an average of(72.3±6.9) years old. The patients were randomly divided into three groups according to random number table, 30 patients in each group. In group A, there were 11 males and 9 females, aged from 61 to 84 years old with an average of (71.60±6.04) years old, body mass index was (26.04±1.95) kg/m², hemoglobin was(128.57±5.98) g/L, treated by 7.5 mg/kg tranexamic acid was injected intravenously before upper tourniquet, and 7.5 mg/kg tranexamic acid after closure of incision and before loosening tourniquet, meanwhile combined with flexion position of hip for 30° to 45° and flexion position of knee for 60° for 70°. In group B, there were 10 males and 20 females aged from 61 to 85 years old with an average of (72.04±7.47) years, body mass index was (25.92±1.70) kg/m², hemoglobin was (127.58±4.37) g/L, treated by 15 mg/kg tranexamic acid injected intravenously before loosening tourniquet. In group C, there were 12 males and 18 females aged from 62 to 85 years old with an average of (73.23±7.36) years, body mass index was (26.07±1.49) kg/m², hemoglobin was (128.31±5.61) g/L, treated with the same amount of normal saline before loosening tourniquet. Intraoperative bleeding volume, postoperative drainage volume, recessive blood loss, total blood loss volume, blood transfusion cases, activated partial thromboplastin time(APTT), prothrombin time(PT), prothrombin international standardized ratio (PT-INR) and indexes of D-dimer(D-D) were compared among three groups, as well as postoperative deep venous thrombosis and pulmonary embolism were observed among three groups.No incision infection occurred in all 90 patients, and all patients were followed up from 4 to 8 months with an average of 6 months without pulmonary embolism occurred. There was no statistical difference in itraoperative bleeding volume among three groups(F=0.299, =0.742), while there were significant differences in postoperative drainage volume, recessive blood loss, and total blood loss among three groups. The number of blood transfusion were as following, 2 cases in group A, 8 cases in group B, and 16 cases in group C, there were statistically significant differences among three groups(χ² =16.01, <0.001). There were no differences in APTT, PT, PT-INR and D-D after operation among three groups(>0.05), and no difference in occurrence of lower limb vein thrombosis after operation.The method of using tranexamic acid before upper tourniquet, after closure of incision and before loosening tourniquet-combined with the flexion position of hip and knee could effectively reduce postoperative drainage volume, recessive bleeding, total blood loss and blood transfusion cases after total knee arthroplasty, while it does not increase risk of deep vein thrombosis and pulmonary embolism.