Ninety-three electroencephalograms (E.E.G.s) were recorded within a week of cardiac resuscitation from 41 patients in whom the subsequent outcome was known to be either recovery of cerebral function or death with associated pathological evidence of gross anoxic brain damage. A statistical analysis of observations on these E.E.G.s yielded a discriminant function for predicting death or survival. Predictions based on each of the 93 individual E.E.G.s would have been correct in 92 and at a confidence level better than 99%. The same discriminant function was found to be applicable to a further 19 patients who died but did not undergo neuropathological studies and to 33 others in whom the clinical picture was complicated by such factors as uraemia or head injury. Thus it seems that the presence or absence of fatal brain damage after cardiac arrest can be reliably predicted from E.E.G.s taken within a week of resuscitation. An estimate of the probability of survival is now routinely included in the clinical report on each E.E.G. taken after cardiac arrest.