&NA;Perturbations in cerebral hemodynamics at the time of ablation of an arteriovenous shunt have been regarded as important in the pathogenesis of swelling and hemorrhage complicating resection of arteriovenous malformations (AVMs). A carotid‐jugular fistula model in the rat had previously been investigated and found to simulate in part the nonhemorrhagic pathophysiology of a large cerebral arteriovenous fistula. Utilizing this model and measuring cerebral blood flow in 14 regions with a [14C]iodoantipyrine autoradiographic technique, the effects of hypocapnea on the cerebral circulation in opened and closed fistulas were investigated. Regional cerebral blood flow (rCBF) in control animals ranged from a median of 53 to 64 ml/100 g/min at a partial arterial carbon dioxide pressure (PaCO2) of 28 ± 2 mm Hg and 85 to 112 ml/100 g/min at a PaCO2of 40 ± 5 mm Hg. In animals with an open carotid‐jugular fistula created 12 weeks before the study, these median rCBF values at comparable PaCO2levels ranged, respectively, from 15 to 39 ml/100 g/min and 50 to 68 ml/100 g/min (the 25th percentile for the open fistula in the hypocapneic group was 15 ml/100 g/min in 5 of the 14 regions studied). In contrast, median rCBF in the closed fistula group ranged from 73 to 100 ml/100 g/min in hypocapneic animals and from 118 to 187 ml/100 g/min in normocapneic animals. These results demonstrate the preservation of CO2reactivity; hypoperfusion in the presence of a carotid‐jugular fistula, hyperemia on fistula occlusion, and the potential to induce cerebral ischemia with hyperventilation in this model of a cerebral arteriovenous fistula. These data caution against the use of hyperventilation during surgery in patients with cerebral arteriovenous malformations. (Neurosurgery25:606‐612, 1989)