The object of this study was to determine whether the addition of information on brain stem reflexes improves the prognostic precision of the Glasgow coma scale for patients with severe head trauma. The study is based on 109 patients with a Glasgow coma score of 7 or less during the first 24 hours after injury. The average age was 23 years. The patients were classified into three groups according to their actual outcome after 6 months: dead, 44 patients; persistent vegetative state and severe disability, 13 patients; moderate disability and good recovery, 52 patients. We then compared, by means of multiple group logistic regression, the prognostic ability of motor responses alone using the Glasgow criteria and of brain stem reflexes via an original approach. We showed that the predictive capabilities of brain stem reflexes were greater than those of motor responses. Although closely related (r = 0.68), the use of these two parameters in a single scale, the Glasgow-Liege scale, improves the precision of prognosis, especially for those head trauma patients with initial and complete loss of consciousness. Age was also revealed to be an important factor for outcome prediction.