Hypothermia has been demonstrated to protect the brain from ischemia or traumatic brain injury. Achieving profound hypothermia has relied on techniques requiring total body cooling, which may result in serious cardiovascular and pulmonary complications. A technique to selectively cool the brain could conceivably exert a marked protection on cerebral structures and provide a relatively bloodless operative surgical field without systemic complications. Accordingly, this approach was tried in 7 rhesus monkeys after induction of general anesthesia. The right internal carotid artery and both internal jugular veins were each occlusively cannulated and connected to a circulation pump. The left internal carotid artery, both external carotid arteries, and both external jugular veins were temporarily clamped to establish severe cerebral ischemia. Using a closed-circuit system, cooled Ringer's lactate liquid (4 degrees C) was infused through right internal carotid artery with outflow draining though both internal jugular veins. Cooled perfusate decreased cerebral temperature to the target temperature of 15 degrees C. Thereafter, pump flow was discontinued, and brains were rewarmed spontaneously, while the temporarily clamped carotid arteries and jugular veins were opened to resume normal cerebral blood circulation. Neurological functions were recorded daily and cerebral histology was examined at the conclusion of the experiment. Magnetic resonance (MR) scans were routinely taken before and 3 weeks after ischemia. In the normothermia control group of five rhesus monkeys, Ringer's solution at 37 degrees C was infused in the same manner as the cold solution with cerebral temperature maintained at 36.7 +/- 0.32 degrees C. Right cerebral temperature decreased from 36.5 +/- 0.49 to 15.5 +/- 2.29 degrees C, and simultaneously the left cerebral temperature decreased from 36.4 +/- 0.38 to 16.3 +/- 2.4 degrees C for 62.8 +/- 9.76 min during selective cerebral cooled Ringer's liquid perfusion. In contrast, rectal temperature was only reduced to 32.4 +/- 0.96 degrees C from a baseline of 37.2 +/- 0.76 degrees C. Internal jugular vein hematocrit was 38.2 +/- 0.31% before perfusion and 2.82 +/- 0.46% at the end of perfusion in profound hypothermia group; hematocrit was 39.7 +/- 0.62% before perfusion and 3.42 +/- 0.38% at the end of perfusion in the normothermia group. In the hypothermic group, neurological functions were normal during 6 months of follow-up, and microscopic examination of brain tissue did not show evidence of pathological changes in hippocampus or medulla. MR scans did not show any cerebral infarction. In contrast, none of the monkeys in normothermia group survived for more than several hours, and microscopic examination of the brain revealed extensive neuronal necrosis within the medulla. Selective cerebral profound hypothermia provides significant histologic and neurologic protection after severe cerebral ischemia. In addition, there were no major complications, and the operative field remained relatively bloodless in the profound hypothermic group.