Recent studies attempting to define the outcome from severe head injury have implied, directly or indirectly, that the severity of injury (as determined by the Glasgow Coma Scale (GSC)) is the sole determinant of outcome. Little attention has been focused on the type of lesion that causes the low GCS score, and there exists an unstated hypothesis that the lesion type is not an important determinant of outcome. No attempt has been made to determine whether patients who have the same GCS score caused by different lesions have the same or different outcomes. Since this is impossible to test without a large number of cases, data were obtained from seven head-injury centers on patients who fulfilled the Glasgow criteria for severe head injury (GCS ≤ 8 for at least 6 hours). Patients were categorized according to a simple classification system comprising seven lesion types, each of which was further subdivided into two GCS score ranges (3 to 5 and 6 to 8). Of 1107 patients, the overall mortality was 41%, but ranged from 9% to 74% among the different lesion categories. Conversely, 26% had good recovery (at 3 months), but among the different lesion groups the range was 6% to 68%. Acute subdural hematoma with GCS scores of 3 to 5 was uniformly the worst problem (74% mortality and 8% good recovery), whereas diffuse injury coma of 6 to 24 hours with GCS scores of 6 to 8 had 9% mortality and 68% incidence of good recovery. Results of this study demonstrate marked heterogeneity within this severe head-injury group and point out that patients with the same GCS score have markedly different outcomes, depending on the causative lesion. The type of lesion is thus as important a factor in determining outcome as is the GCS score, and both must be considered when describing severely head-injured patients.