To evaluate the efficacy of mild hypothermia for the treatment of severe subarachnoid hemorrhage (SAH), we analyzed the outcome of 89 patients who suffered SAH between 1995 and 1999. The patients were divided to 2 groups. Fifty patients who were treated before June 1997 constitute the control group because the cerebral hypothermia was not performed in our hospital in this period. The other 39 patients constitute the hypothermia group. Mild hypothermia therapy was performed for 12 of the hypothermia group, 10 underwent direct surgery and the remaining 2 were treated conservatively. A target core body temperature of 33-34 degrees C was obtained using a cooling blanket. We aimed at intracranial pressure under 20 mmHg. Eight of the 12 patients treated with hypothermia underwent external decompression, and 7 were also administered thiamyral Na. The clinical outcome was GR in 1, MD in 4, SD in 1, VS in 3, and Dead in 1 among the 10 patients in the hypothermia group who underwent surgery. Two conservatively treated patients died. The outcome of patients with WFNS grade 1,2,3, and 5 in both groups did not differ statistically. The outcome of WFNS grade 4 patients of the hypothermia group was statistically better than the control group. Symptomatic vasospasam was observed in 40% of the hypothermia patients. The complications of mild hypothermia was observed in all 12 patients, including 10 pneumonia, 8 hyperpottacemia, 8 liver dysfunction and 4 dysfunction of coagulation. Mild hypothermia therapy could be continued in all patients in spite of these complications. Our study suggests that mild hypothermia for the treatment of severe SAH is effective for controlling ICP and preventing cerebral ischemia, but vasospasm cannot be prevented.