The Glasgow Outcome Scale (GOS), two decades after its description, remains the most widely used method of analyzing outcome in series of severely head-injured patients. This review considers limitations recognized in the use of the GOS and discusses a new approach to assessment, using a structured questionnaire-based interview. Assignments can be made to an extended eight-point scale (GOSE) as well as the original five-point approach-in each case, with a high degree of interobserver consistency. The assignments are coherent with the principles of the World Health Organization classification of impairments, disabilities, and handicaps, and their validity is supported by strong associations with the results of neuropsychological testing and assessment of general health status. The need to allow for disability existing before injury, issues concerning the time of assessment after injury, and subdivisions of the scale into "favorable" and "unfavorable" categories are discussed. It is concluded that, in its improved structured format, the Glasgow Outcome Scale should remain the primary method of assessing outcome in trials of the management of severe head injury.