Adequacy of transfusion based exclusively on clinical observations of hemodynamic changes and quantities of blood lost during total hip arthroplasty (THA) in 15 adults was tested by serial measurements of 51-Cr red-cell volume (RCV), 125-I RISA plasma volume (PV), and peripheral hematocrit (Hv) determined immediately before and 2 and 48 hours after operation. Blood transfusion and intravenous-fluid therapy were based on clinical observations and designed to restore blood volume (BV) to within 10 percent of its preoperative level prior to the isotopic determination 2 hours postoperatively. (When that determination showed a BV less than 90 percent, transfusions were given to restore it to between 90 and 100 percent). Isotopic data 2 hours after operation showed that the BV reduction--thought clinically to be less than 10 percent in every case--averaged 21.6 percent (range: 6 to 39 percent) and was shared by RCV and PV. Isotopic measurements 48 hours after operation showed that patients whose BV at 2 hours was within 10 percent of the preoperative BV were able to compensate for an average continuing RCV reduction of 23.9 percent by expansion of PV, resulting in an average BV reduction of 11.4 percent. Hv was a valuable guide for transfusion therapy at 48 hours, but not 2 hours after operation. In view of inadequacy of blood replacement based on clinical observations and the magnitude and variability of the 2 hour postoperative isotopically measured BV reductions in our patients, we recommend obtaining a derived BV by measuring PV and Hv preoperatively and 2 hours postoperatively as a further guide to transfusion therapy for each patient under-going THA. This is simpler and more practical than the two-isotope technique and is more sensitive than clinical observation alone. If clinical observation had been supplemented with a derived BV in our 15 cases, 80 percent instead of 30 percent would have had their BV restored to within 10 percent of the preoperative BV.