The research examined the effects of alcohol and other variables on cognitive outcome after severe head injury. Alcohol consumption habitually and at the time of injury were strongly related, and both were related to age and educational level but not injury severity. Covariance analysis to remove the effects of age and education showed a reduction in the main effects, so that only alcohol consumption at injury was a significant predictor of memory, but not other cognitive areas late after injury. There were significant interactions between severity of injury (post-traumatic amnesia), and alcohol habitually or at time of injury, with increasing alcohol consumption increasing the size of the memory deficit. To have a short post-traumatic amnesia and have drunk heavily led to a worse memory score than that found in patients with a considerably longer post-traumatic amnesia who had drunk lightly or not at all.