Objective To analyze the impact of ivaroxaban on hidden blood loss and blood transfusion rate after primary total knee arthroplasty (TKA) by comparing with the use of low molecular weight heparin. Methods Between December 2009 and January 2011, the clinical data from 90 patients undergoing primary TKA were retrospectively analyzed. At 12 hours after operation, 45 patients were given ivaroxaban (10 mg/d) in the trial group and low molecular weight heparin injection (0.4 mL/d) in the control group for 14 days, respectively. There was no significant difference in gender, age, disease duration, or range of motion between 2 groups (P gt; 0.05). Results The operation time was (92.32 ± 23.13) minutes in the trial group and (89.81 ± 18.65) minutes in the control group, showing no significant difference (t=0.26, P=0.79). The hidden blood loss was (40.18 ± 14.85) g/L in the trial group and (34.04 ± 12.96) g/L in the control group, showing significant difference (t=2.09, P=0.00); the dominant blood loss was (30.60 ± 2.89) g/L and (28.85 ± 8.10) g/L respectively, showing no significant difference (t= 1.37, P=0.17). The blood transfusion rate was 73.33% (33/45) in the trial group and 55.56% (25/45) in the control group, showing no sigificant difference (χ2=3.10, P=0.08); the transfusion volume was (1.44 ± 1.09) U and (1.06 ± 1.17) U respectively, showing no significant difference (t=1.58, P=0.11). Stress ulcer occurred in 1 case of the trial group; symptomatic deep vein thrombosis of lower extremity and asymptomatic muscular venous thrombosis developed in 1 case and 4 cases of the control group respectively. Conclusion Ivaroxaban has effect on the hidden blood loss after primary TKA, which may increase postoperative blood loss and blood transfusion rate. The changes in hemoglobin should be monitored during the anticoagulant therapy, and the blood volume should be added promptly.