BACKGROUND: From 1994, we have used therapeutic hypothermia in patients with severe traumatic brain injury (Glasgow Coma Scale scores of 5 or less). In 2000, we altered the target temperature to 35 degrees C from the former 33 degrees C, as our findings suggested that cooling to 35 degrees C is sufficient to control intracranial hypertension, and that hypothermia below 35 degrees C may predispose patients to persistent cumulative oxygen debt. We attempted to clarify whether 35 degrees C hypothermia has the same effect as 33 degrees C hypothermia in reducing intracranial hypertension and whether it is associated with fewer complications and improved outcomes. METHODS: We compared intracranial pressure (ICP) and biochemical parameters in the 30 patients treated with 35 degrees C hypothermia (January 2000 to June 2005) with those in the 31 patients treated with 33 degrees C hypothermia (July 1994 to December 1999). RESULTS: Patient characteristics were similar in the two groups. The mean temperature during hypothermia was 35.1 +/- 0.7 degrees C in the 35 degrees C hypothermia group and 33.4 +/- 0.8 degrees C in the 33 degrees C hypothermia group. Mean ICP was controlled under 20 mm Hg during hypothermia in both the 35 degrees C hypothermia and 33 degrees C hypothermia groups. The incidence of intracranial hypertension and low cerebral perfusion pressure did not differ between the two groups. The 35 degrees C hypothermic patients exhibited a significant improvement in the decline of serum potassium concentrations during hypothermia and in the increment of C-reactive protein after rewarming. The mortality rate and the incidence of systemic complications tended to be lower in the 35 degrees C group. CONCLUSIONS: Cooling patients to 35 degrees C is safe and the ICP reduction effects of 35 degrees C hypothermia are similar to those of 33 degrees C hypothermia.