To compare the effect of long-term mild hypothermia versus short-term mild hypothermia on the outcome of 215 severe traumatic brain injured patients with cerebral contusion and intracranial hypertension. At three medical centers, 215 patients aged 18 to 45 years old with an admission Glasgow Coma Scale < or =8 within 4 h after injury were randomly divided into two groups: long-term mild hypothermia group (n = 108) for 5+/-1.3 days mild hypothermia therapy and short-term mild hypothermia group (n = 107) for 2+/-0.6 days mild hypothermia therapy. All patients had intracranial hypertension and frontotemporoparietal contusion with midline shift >1 cm confirmed on computed tomographic scan. Glasgow Outcome Scale at 6-month follow-up, 47 cases had favorable outcome (43.5%), and other 61 cases had unfavorable outcome (56.5%) in the long-term mild hypothermia group. However, only 31 cases had favorable outcome (29.0%), and other 76 cases had unfavorable outcome (71.0%) in the short-term mild hypothermia group (P < 0.05). The intracranial pressure significantly rebounded after rewarming in the short-term mild hypothermia group, but not in the long-term mild hypothermia (P < 0.05). Furthermore, the incidence of stress ulcer, epilepsy, pulmonary infection, intracranial infection did not significantly differ between the two groups (P > 0.05). Compared with short-term mild hypothermia, long-term mild hypothermia significantly improves the outcome of severe traumatic brain injured patients with cerebral contusion and intracranial hypertension without significant complications. Our data suggest that 5 days of long-term cooling is more efficacious than 2 days of short-term cooling when mild hypothermia is used to control refractory intracranial hypertension in patients with severe traumatic brain injury.