Major advances have been made in understanding the pathophysiological events after severe human traumatic brain injury, and consequently, many compounds have been tested in clinical trials. Thus far, no Phase III trials have been clearly successful, in human neurotrauma, although several Phase II studies have shown apparent benefit. This review is an attempt to identify factors that could be responsible for some of these failures. Recommendations are made that attempt to avoid these pitfalls in the future. Five criteria for future conduct of clinical trials are proposed. The usefulness of animal models for traumatic brain injury and their ability are discussed. Clearly, it is now becoming accepted that mechanism-driven trials, in which individual pathophysiological mechanisms are targeted, may be preferable in this heterogeneous patient population. The degree of brain penetration, the safety and tolerability of the compound, and end points used for outcome assessment are major influences upon the success of these trials. New approaches in developing, conducting, and analyzing these clinical trials should be considered in the future, if the costly failures of the past are not to be repeated, with the advent of newer "neuroprotective agents" and techniques.