Summary Analyses have been performed upon data derived from 503 patients having coma and apnea in the Collaborative Study of Cerebral Survival and from subsets of 44 survivors and 102 cases in which details of temporal course and outcome were specified. Three methods of analysis were employed. The first consisted of correlation of proposed clinical signs with a series of presumptive definitions of brain death, the most rigorous of which included pathologic diagnosis. The second dealt with a conservative definition of brain death based upon the condition of electrocerebral silence with comparison of time course of clinical signs in patients having presumptive brain death versus that in patients who survived. Third, discrimination value of individual signs and combinations was evaluated by phi coefficients in relationship of presumptive brain death versus cases without evidence of brain death. These analyses lead to the exclusion of some of the proposed clinical signs. Body temperature and blood pressure showed no correlational trend and had no discriminative value. The spinal reflexes showed high persistence rates in cases of presumptive brain death and also had low phi coefficients. Some reflexes, the audio‐ocular and abdominal, were excessively sensitive, being commonly absent both in brain death as well as in non‐brain death cases and survivors. The snout and jaw reflexes were the two cephalic reflexes most likely to persist in presumptive brain death and to have low phi coefficients. The clinical signs recommended as diagnostic criteria include: coma, apnea, and the absence of the light, vestibular, oculocephalic and corneal reflexes. These criteria existed in the absence of severe hypothermia (below 90|Mo F) or drug intoxication. The four cephalic reflexes were identified on the basis of prevailing absence in presumptive brain death, return in cases ultimately surviving, and high phi coefficients. Spontaneous movements are of borderline clinical value but may be retained in view of lay interpretations. Pupillary dilatation is not a criterion, except in the cases of brain death due to cardiac arrest. The pharyngeal, swallow and cough reflexes need not be criteria for brain death, but are of value in the sequential identification of drug‐intoxicated survivors. EEG recording is necessary in addition to the clinical criteria in order to avoid incorrect identification of brain death in patients with primary brain‐stem disorders and in potential survivors.