The cerebral effects of alterations in plasma osmolality (Osm) and colloid oncotic pressure (COP) were examined in normocarbic, normothermic, pentobarbital-anesthetized rabbits that had been subjected to cryogenic brain injury. Monitored variables in all animals included mean arterial, right atrial, and intracranial pressures (MAP, CVP, and ICP), electroencephalographic (EEG) recordings, and cerebral blood flow (CBF). When surgical preparation was complete, a left parietallesion was produced with liquid nitrogen. Group 1 (control, n = 8) animals subsequently received only maintenance fluids [lactated Ringer's solution (LR)]. One hour after injury, 3 other groups of animals underwent 45 minutes of plasmapheresis, carried out by arterial phlebotomy (packed red cells returned), with separated plasma being replaced by one of three fluids given in amounts sufficient to maintain MAP and CVP at baseline values. The three fluids were 1) 6% hetastarch in hypo-osmotic LR [Group 2 (Hypo-Osm), n = 6; COP = 21 mm Hg, Osm = 130 mOsm/kg]; 2) iso-osmotic LR [Group 3 (Hypo-COP, n = 8; COP = 0; Osm = 305]; and 3) 6% hetastarch in iso-osmotic LR [Group 4 (Iso-Osm/COP), n = 8; COP = 21, Osm = 310]. The animals were killed by exsanguination 25 minutes after completion of plasmapheresis. The brain was removed, the hemispheres separated, weighed and sliced, and the specific gravities (SpGr) of the regional tissue determined. There were no differences in MAP, CVP, regional CGF, or EEG activity among the groups. In Group 2, osmolality decreased by 12.7 ± 2.9 mOsm/kg (mean ± SD, COP unchanged), while ICP increased to 22 ± 6 mm Hg (P < 0.001 vs. Groups 1, 3, and 4) and CBF decreased to 73% of that recorded prior to plasmapheresis. In Group 3, COP was reduced by 10.8 ± 2.2 mm Hg (Osm unchanged), while both COP and osmolality were unchanged in Group 4. ICP increased in these groups to 9 ± 3 and 13 ± 4 mm Hg, respectively, compared with 14 ± 9 in Group 1. The lesion itself caused decreases in regional SpGr in all animals (as compared with the contralateral hemispheres). The only significant fluid-related difference, however, was associated with a reduction in osmolality - and only in tissues remote from the injury. Changes in COP had no detectable effects. These results indicate that in the acute phase of brain injury, reductions in COP do not play an important role in brain edema formation, and that changes in osmolality alter water content only in relatively normal brain regions.