Previous studies have shown that diaspirin cross-linked hemoglobin (DCLHb(TM), 10 g/dL) decreases cerebral ischemia and the resultant injury in a dose-dependent manner, requiring large volumes of DCLHb(TM) for maximum efficacy. We assessed the effect of a more concentrated (20 g/dL) and more hyperoncotic preparation of DCLHb(TM) on cerebral infarction volume. Immediately after middle cerebral artery occlusion, rats were randomized to one of the following groups, Control, hematocrit not manipulated; 10/Hb, hematocrit decreased to 30% with 10% DCLHb(TM) (oncotic pressure 43 mm Hg); 7.5/Alb, hematocrit decreased to 30% with 7.5% albumin (oncotic pressure 43 mm Hg); 20/Hb, the same dose of DCLHb(TM) (20%, oncotic pressure 129 mm Hg) as the 10/Hb group (half the volume); or 15/Alb, the same dose of albumin (15%, oncotic pressure 130 mm Hg) as the 7.5/Alb group (half thr volume). After 90 min of ischemia, 72 h of reperfusion was allowed. Infarction volume (mm(3), mean +/- SD) was less in the DCLHb(TM) groups (10/Hb = 79 +/- 17; 20/Hb = 51 +/- 14) than the oncotically matched albumin groups (7.5/Alb = 124 +/- 21; 15/Alb = 85 +/- 18) and the Control group (135 +/- 17) (P < 0.05). These data indicate that in this model of cerebral ischemia, DCLHb(TM) decreases ischemic brain injury more effectively than albumin, and that a hyperoncotic preparation of DCLHb(TM) is preferable.