Paraplegia from spinal cord ischemia during thoracoabdominal aneurysm repair remains an unpredictable and unpreventable complication. In an effort to prevent spinal cord ischemia during aortic cross-clamping, preoperative angiographic localization of the blood supply to the spinal cord was performed in dogs. Sixteen animals underwent 60 minutes of thoracoabdominal aortic cross-clamping either without (control, n = 8) or with (shunted, n = 8) a selective shunt. Shunting was performed from the aortic arch to that isolated aortic segment angiographically shown to supply the thoracolumbar anterior spinal artery. Spinal cord blood flow was measured with microspheres just prior to cross-clamping, at 5 and 60 minutes after cross-clamping and at 5 minutes after restoration of aortic blood flow. Functional neurologic outcome was evaluated in animals at 24 hours postoperatively. Shunting did not decrease spinal cord injury. Seven of the 8 animals in the control group and 7 of the 8 in the shunted group developed paraplegia or paraparesis. Thoracic, but not lumbar spinal cord blood flow, was significantly increased in shunted animals. Spinal cord blood supply in dogs may be more segmental than previously believed. Technical problems in angiographic localization, spinal artery spasm, loss of spinal cord autoregulation or poor collateral circulation from the distal thoracic to the lumbar cord may also account for these results. Although shunting to aortic segments supplying the anterior spinal artery during thoracoabdominal aortic clamping may be attractive in humans, no benefit could be shown in this experimental model.