Traumatic brain injury initiates several metabolic processes that can increase the primary injury. It is well established that in severe head injuries, posttraumatic secondary insults, such as brain hypoxia, hypotension or anemia, exacerbate neuronal injury and lead to a poorer outcome. Experimental and clinical evidence suggests that moderate hypothermia (32-34 degrees C), may limit some of these deleterious secondary metabolic responses. Recent laboratory studies and prospective controlled clinical trials of induced moderate hypothermia for relatively short periods (24-48 h) in patients with severe head injury, have demonstrated good intracranial pressure control and better outcome when compared with patients maintained in normothermia and given conventional treatment. Despite its proven clinical role in neuroprotection, hypothermia research has been inconstantly followed for various reasons. In this paper we review the mechanisms of neuroprotection in hypothermia, the different preclinical and clinical studies that favor its use as a neuroprotector in severe head injury or in patients in whom high intracranial pressure is refractory to first tier measures. The evidence that favors hypothermia is discussed. We also discuss the negative results of the still unpublished multicentre trial on prophylactic moderate hypothermia developed in the USA. The main problem with moderate hypothermia is the lack of a systematic methodology to induce and maintain it. Also, optimal duration of its use and the methodology and timing for rewarming have not been determined. Consequently, the results of different trials are difficult to analyze and compare. However, most evidence suggests that hypothermia provides remarkable protection against the adverse effects of neuronal damage that is exacerbated by secondary injury. Further prospective controlled trials with clearly defined methodology are needed before this method is implemented in daily clinical practice. The most important task for the years to come may be to focus on refining this procedure, defining the optimal time of cooling and rewarming and to optimize the methods of rapidly inducing and maintaining low temperature. It is also essential to define the most appropriate method and velocity of the rewarming phase, in which many successfully controlled patients deteriorate and die.