Blood conservation has been most successful when blood salvage techniques have been combined with postoperative normovolemic hemodilution. The hemodynamic and myocardial metabolic responses to normovolemic hemodilution were assessed in a prospective randomized trial. Twenty-seven patients were randomized to receive either blood and colloid solutions (colloid group, 13 patients) or crystalloid fluids (crystalloid group, 14 patients) following elective coronary revascularization. Although seven patients in the crystalloid group received blood products when the hemoglobin level fell below 7 gm/dl, blood bank requirements were less in the crystalloid group (colloid, 3.6 +/- 1.2 L; crystalloid, 1.5 +/- 1.0 L, p less than 0.01). The crystalloid group received twice as much fluid to maintain normovolemia (left atrial pressure between 8 and 10 mm Hg) in the first 72 hours postoperatively (colloid, 6.5 +/- 1.9 L; crystalloid, 14.5 +/- 3.1 L, p less than 0.01). The infusion of large volumes of crystalloid fluids resulted in a progressive postoperative anemia (hemoglobin: colloid, 12.1 +/- 1.6 gm/dl, crystalloid 8.9 +/- 1.7 gm/dl, p less than 0.01, 20 hours postoperatively). Although the crystalloid-treated patients had peripheral edema, pulmonary edema could not be documented and there was no difference in the physiological shunt fractions between the two groups. Preload (left atrial pressure), afterload (mean arterial pressure), and cardiac index were similar in the two groups. The crystalloid group had a delayed recovery of myocardial oxygen and lactate extraction postoperatively. Volume loading and atrial pacing 3 to 5 hours postoperatively maintained myocardial lactate extraction in the colloid group but decreased myocardial lactate extraction to ischemic levels in the crystalloid group. The use of crystalloid rather than colloid fluids in the early postoperative period conserved blood products but resulted in postoperative anemia and was associated with a delay in myocardial metabolic recovery. Normovolemic hemodilution should be employed with caution in patients who are at risk of perioperative ischemic injury.