Experimental hemodilutional therapy has been shown to raise collateral perfusion to acutely ischemic brain regions distal to occluded internal carotid (ICA) and middle cerebral (MCA) arteries and to reduce infarct size. Superficial temporal (STA)-MCA anastomosis surgically establishes additional regional collateralization, and this bypass angiographically enlarges over time. Despite bypass patency verification by Doppler recording made at the edge of the craniectomy, the microsurgical STA-MCA anastomosis in 11 stroke patients did not produce early changes in cerebral perfusion parameters in the MCA territory of either hemisphere as determined by 133xenon inhalation. Therefore, hemodilution was initiated in an effort to increase cerebral perfusion during this marginal period when the STA was beginning to dilate progressively. Incremental venesections with equal intravenous volume replacement with 5% human serum albumin caused a significant lowering of the hematocrit from 40 ± 1 to 33 ± 1%. This isovolemic hemodilutional therapy resulted in significant mean regional cerebral blood flow (rCBF) elevations of 23 ± 5% (SE) in the bypassed MCA territory and of 25 ± 6% in the opposite MCA region. The mean gray flow (F1) in the involved and homologous MCA regions significantly increased 27 ± 8% and 30 ± 11%, respectively. Similarly, the initial slope index (ISI2) significantly rose by 17 ± 5% in the bypassed MCA territory and by 18 ± 6% in the homologous region. These data objectively support the premise that reductions in hematocrit without intravascular volume expansion augment cerebral blood flow, probably by reducing blood viscosity. Moreover, inovolemic hemodilution, the effect of which lasts several weeks, may be useful adjunct therapy for the postoperative care of patients with marginal cerebral circulation who undergo cerebral revascularization.