The quality of outcome after severe closed head injury has become of increasing concern to neurosurgeons. The assessment of residual deficits in patients who have recovered from closed head injury can be very difficult. Many patients are classified as having a good recovery according to the Glasgow Outcome Scale (GOS), but this may be insufficiently focused or sensitive to demonstrate mental deficits objectively. We investigated 33 patients with severe closed head injury who subsequently were diagnosed as having made a good recovery according to the GOS. The severity of the injury was determined by the Glasgow Coma Scale (GCS) and by the presence of a midline shift in the preoperative CT scans. There was a minimal interval of 15 months (x= 1080-5 days, SD = 491 days) between injury and time of neuropsychological testing. Their performance was compared with that of 15 orthopacdic cases. Residual neuropsychological deficits can be demonstrated on the majority of measures in a group of patients who have achieved good recovery on the GOS. Midline shift in preoperative CT scans is not of prognostic value for long-lasting neuropsychological deficits.