The present study was designed to investigate the outcome of 100 surviving patients who were admitted consecutively to a regional neurosurgical unit with a diagnosis of either moderate or severe TBI. An additional group of 44 SAH patients was recruited to enable tentative comparisons to be drawn between groups with acute brain injury arising from these different causes. At intervals of 6, 12 and 24 months post-injury, global outcome was estimated using the Glasgow Outcome Scale (GOS) and patients were assessed using a battery of neuropsychological tests. Information concerning the extent of the patient's recovery was also collected at each time point. Poorer GOS outcome was associated with greater severity of initial injury. However, complications arising at the acute stage of treatment also influenced outcome. Because of the limited sensitivity of the GOS, a more detailed outcome scale, the Edinburgh Extended Glasgow Outcome Scale (EEGOS), was devised and applied. Results of the neuropsychological assessment in the TBI patient groups showed that severe cognitive impairment was typically associated with more severe initial injury. These results are in agreement with those of Rimel et al. (1982) in suggesting that patients with moderate TBI experience a level of morbidity intermediate between those with severe and those with minor TBI, and contrast with those of Anderson et al. (1994) for reasons which are discussed. Despite the relationship between severity of injury and cognitive impairment, functional problems were reported to a similar extent by relatives of both the severe and moderate TBI patients. The results suggest that the brain injury impacts on everyday life to a greater extent and in many more patients than might be expected on the basis of formal cognitive assessment. Comparisons between the TBI and SAH groups were heavily influenced by a difference in the estimated severity of initial brain injury, but there were similarities in outcome between groups. Formal assessment of cognitive function is useful and informative in the conditions studied, but that other measures provide important supplementary information about the ways in which acute brain injury influences the lives of patients.