Background Total knee arthroplasty (TKA) is an important therapy for the treatment of various late-stage knee diseases. However, it has been observed that patients have lower hemoglobin (HB) counts postoperatively, which are significantly inconsistent with the measured blood loss. Although the concept of hidden blood loss has been presented in 2000, very little clinical attention has been paid since then. In this study, we investigated the characteristics and influential factor of hidden blood loss after TKA in treating knee osteoarthritis. Methods One hundred and eight patients, with 42 males and 66 females, were retrospectively analyzed. The perioperative blood loss and the hidden blood loss following TKA were calculated by Gross formula. Potential factors affecting perioperative and hidden blood loss included gender, surgical time, tourniquet time, hemostasis (during operation with deflating tourniquet), and reinfusion of salvaged blood. Results The average perioperative blood loss was found to be (1553 +/- 356) ml and the average hidden blood loss was (793 +/- 223) ml. The hidden blood loss of the male was significantly higher than that of the female, and was reduced by hemostasis during the operation with a deflating tourniquet. Hidden blood loss was not affected by postoperative blood salvage and reinfusion, surgical time, or tourniquet time. Conclusions When calculating the value of hidden blood loss by Gross formula, the lowest value of hematocrit should be chosen as the parameter so that hidden blood loss would not be underestimated. No significant differences could be found in hidden blood loss for males compared to females. Hidden blood loss can be reduced by hemostasis during operation with a deflating tourniquet. Although reinfusion of salvaged blood could not reduce the perioperative blood loss or hidden blood loss, it was still an effective method to reduce transfusion rate. Chin Med J 2011;124(11):1653-1656