Acute normovolemic hemodilution (ANH) involves withdrawal of whole blood with concurrent infusion of fluids to maintain normovolemia. The aim of this study was to quantify the efficacy and safety of preoperative ANH with a systematic review and meta-analysis. Randomized controlled trials were identified through MEDLINE (1966-2002) and the Cochrane Controlled Trials Database and with hand searching of journals. All trials of preoperative ANH reporting on allogeneic transfusion, bleeding, or adverse outcomes were included. Paired reviewers independently abstracted data. Outcomes were pooled using random-effects models. A total of 42 trials compared hemodilution to usual care or to another blood conservation method. The risk of allogeneic transfusion was similar among patients receiving ANH and those receiving usual care (relative risk [RR], 0.96; 95% CI, 0.90-1.01), or another blood conservation method (RR, 1.11; 95% CI, 0.96-1.28). Hemodiluted patients, however, were transfused from 1 to 2 fewer units of allogeneic blood. They had less total bleeding than patients receiving usual care (91 mL; 95% CI, 25-158 mL), although more intraoperative bleeding. Only one-third of studies reported on adverse events. The literature supports only modest benefits from preoperative ANH. The safety of the procedure is unproven. Widespread adoption of ANH cannot be encouraged.