BACKGROUND: Almost all of the blood loss occurring postoperatively in total knee arthroplasty (TKA) is due to the use of an intraoperative tourniquet in all cases. So we tried having blood transfusions done, not by anesthesiologists intraoperatively but by orthopedic surgeons postoperatively. The purpose of this study is to analyze postoperative blood loss and transfusion practice in TKAs. METHODS: We analyzed retrospectively the medical records of 64 TKAs in 40 patients between March, 1997 and February, 1999. RESULTS: Six male and fifty-eight female patients were enrolled. Their mean preoperative, immediate postoperative, and post-transfusion hematocrit were 37.1+/- 3.4, 34.4+/- 3.3, and 34.4+/-3.9, respectively. The preoperative and post-transfusion hematocrit were measured in all cases, but immediate postoperative hematocrit was measured in only 21 cases, and the hematocrit of 17 among the 21 cases was over 32 percent. Drainage amount, for the 1st postoperative day were 843 328 ml, and the total drainage amount was 993+/-362 ml. The blood transfusion amount was 2.6+/-0.9 units. There were no statistically significant differences in preoperative hematocrit, immediate postoperative hematocrit, postoperative drainage amount and postoperative transfusion amount between the general and regional anesthetic group; Preoperative hematocrit did not statistically affect the postoperative transfusion amount, but the postoperative drainage amount was statistically associated with an increased postoperative transfusion amount. CONCLUSIONS: Based on these results, it was concluded that postoperative blood transfusion in TKA was to be done on the basis of clinical impressions of orthopedic surgeons in which postoperative drainage amount was the most important consideration. Therefore, postoperative transfusion should be based on appropriate transfusion guidelines and a careful clinical examination of the patient.