Hemodilution, one of several methods proposed to decrease homologous blood transfusion in elective surgery, has not been studied in a prospective controlled trial to determine if it is successful. A prospective, randomized controlled study was conducted to determine if hemodilution can serve as an alternative to preoperative autologous blood donation. Fifty patients were randomized to preoperatively deposit 3 units of autologous blood or to undergo hemodilution immediately before elective radical retropubic prostatectomy. All patients were treated under a standard protocol, including surgery performed by a single surgeon. The preoperative deposit groups received a mean of 2.44 +/- 1.0 units of blood; 2 of 25 patients required homologous blood transfusion for blood loss of 2600 mL and 1700 mL. The hemodilution group received a mean of 2.88 +/- 0.4 units of autologous blood: no hemodilution patient received homologous blood. At discharge, the mean hematocrit for the preoperative deposit group was 35.5 +/- 4.9 (0.35 +/- 0.05), and that for the hemodilution group was 31.8 +/- 4.7 (0.32 +/- 0.05) (p less than 0.001). There were no differences in perioperative morbidity in the treatment groups. The best predictor of discharge hematocrit was the initial hematocrit of the patient. It can be concluded that hemodilution can safely replace or at least augment preoperative autologous donations as a means of decreasing homologous blood transfusion in study patients. These results can be applied to any elective surgery procedure in which a 1000-mL blood loss is anticipated. Other advantages of hemodilution, including convenience, lower cost, and better preservation of all components of autologous blood, suggest that this practice deserves wider application.