When brain death in children occurs, commonly the scalp feels cold despite a normal core temperature. This phenomenon might reflect absent cerebral blood flow and metabolic activity. The authors, therefore, measured rectal-scalp temperature differences in critically ill comatose children to test the hypothesis that a particular temperature difference may correlate with clinical brain death. In a prospective cohort study set in a pediatric intensive care unit, rectal-scalp, rectal-abdomen, and rectal-mastoid temperatures in critically ill comatose children older than 18 months of age were measured before and during brain death evaluations. Twelve children were enrolled. Clinical criteria for brain death were met by seven patients, and five patients survived. All of the seven children who died had rectal-scalp temperature differences greater than 4 degrees C (mean = 6.7, range = 6.0-7.4) at the time of clinical brain death. No survivor had a rectal-scalp temperature difference of 4 degrees C at any time (mean = 3.4, range = 2.9-3.9). Rectal-scalp temperature differences of those who died and those who survived were significantly different at the P < 0.005 level. Rectal-abdomen and rectal-mastoid temperature differences did not correlate with clinical brain death or rectal-scalp temperature difference. In this preliminary study a rectal-scalp temperature difference of greater than 4 degrees C correlates with clinical criteria for brain death in children.