We disagree with Dr. Karkouti’s comment that, when performed for total knee replacement (TKR), hemodilution is not expected to be an effective blood-sparing strategy. Because TKR is associated with extensive postoperative blood loss (which is not controlled by surgical techniques), transfusing blood that is rich in platelets and clotting factors has the theoretical potential to decrease postoperative bleeding and allogeneic blood requirements. Furthermore, when compared with control, normovolemic hemodilution (NVHD) has been shown to significantly reduce allogeneic blood transfusion after TKR (1,2). Similarly, after TKR, tranexamic acid administration significantly reduces allogeneic blood transfusion (3). Because both blood conservation strategies may be associated with unwanted adverse effects, we believe that a comparative study designed to assess the relative efficacy of these two strategies is not only appropriate but essential before evidence-based decisions can be made. Dr. Karkouti suggests that, because NVHD reduces patient hematocrit, our study design increased the likelihood that the NVHD study group would receive allogeneic blood. However, the decrease in hematocrit is not study-design dependent. By definition, acute hemodilution is associated with a decreased hematocrit. Therefore, it would be impossible to compare the relative efficacy of NVHD versus any other blood-sparing protocol without accepting this inherent limitation. Furthermore, despite the hemodilution-induced decrease in hematocrit, when compared with control, NVHD significantly reduced the allogeneic blood requirement after TKR (1). Because the frequency of postoperative transfusion is influenced by an institution-specific “transfusion culture,” predetermined “transfusion trigger,” as well as other patient and doctor variables, expected transfusion rates can be meaningful only if they are the result of vigorously controlled, randomized, prospective studies. Unfortunately, Dr. Karkouti presents “soft” data to support the claim that the hemodilution-induced reduction in perioperative hematocrit resulted in the “extremely frequent” transfusion rate (65%) among our NVHD treatment group. Neither the citation that the “expected transfusion rate for this procedure [TKR] in patients lacking autologous blood is less than 30%” (4), nor the fact that at Dr. Karkouti’s institution, “the current transfusion rate is less than 15%” fulfill the criteria required to facilitate meaningful interpretation. In contrast, in a prospective, randomized study, Olsfanger et al. (1), administered allogeneic blood to 100% of the control patients. Furthermore, in a similarly designed study, Hiippala et al. (3), administered allogeneic blood to 90% of the control patients. Thus, it is incorrect to suggest that NVHD increases the likelihood of allogeneic blood transfusion after TKR. Finally, these transfusion statistics support our comment that “randomizing patients into a treatment group that would most probably be exposed to more allogeneic blood requirements (and blood-induced complications) was ethically unacceptable.” Edna Zohar MD Brian Fredman MB, BCh Martin Ellis MB, BCh Robert Jedeikin MBChB, FFA(SA)