Many compounds have now been tested that were expected to ameliorate the secondary ischemic brain damage after severe head injury. Thus far, none of these have been clearly successful. This review is an attempt to identify factors that could be responsible for some of these failures. Recommendations are made that could help to avoid these pitfalls in the future. The usefulness and criteria for use of animal models for traumatic brain injury to depict human head injury are discussed. Clearly, it has now become widely accepted that mechanism-driven trials, in which individual pathophysiological mechanisms are targeted, are preferable in this heterogeneous patient population. Other factors, such as the effect of brain penetration, safety and tolerability of the compound, and the interface between the pharmaceutical industry and academics are a major influence in the success of these trials. Furthermore, different ways of analyzing trials such as sequential analysis and newer, alternative end points should be considered. Pharmacological agents will never be the "magic bullet" for a process as heterogenous in pathophysiological mechanisms as traumatic brain injury. This does not imply that the role of neuroprotective compounds will not be important in the future. New approaches in developing, conducting and analyzing these expensive clinical trials must be devised in the future.