Selective brain cooling in humans, with venous blood returning from the head surface as the relevant heat sink, was proposed more than two decades ago as a mechanism protecting the brain against damage in hyperthermic conditions. Brain cooling was inferred from decreases of tympanic temperature under the premise that it reflected brain temperature closely, even in conditions of external head cooling. In mammals with a well-developed carotid rete selective brain cooling and its quantitative relevance are experimentally well established by directly monitoring brain temperature. For humans, however, the dispute about the existence and physiological relevance of selective brain cooling has remained unsettled, especially, as far as arguments have been exchanged on the basis of thermophysiological data and model calculations considering brain metabolism, brain hemodynamics and the anatomical preconditions for arterio-venous heat exchange. In this essay two seminal studies in support of the existence of human selective brain cooling in the condition of exercise hyperthermia, with and without dehydration, are re-examined from two points of view: first the stringency of the working hypotheses underlying data evaluation and their subsequent fate. Second the minimum theoretical requirements for data interpretation. The working hypotheses supporting data interpretation in favor of selective brain cooling in humans were heuristic and/or had become questionable at the dates of their application; today, they may be considered as outdated. Data interpretation becomes most conclusive, if tympanic temperature simply is not taken into account.