In order to test the hypothesis that Wernicke's encephalopathy is of topographic rather than of pathogenetic specificity we examined the brains of 49 patients without any evidence of chronic alcoholism. They had died at least four days after an event of severe hypoxia-ischemia. They all showed extensive lesions in the cortex, in the thalamus and in other regions. In 19 of them there was additional necrosis in the mamillary bodies which apparently was of the same age as the associated cortical and thalamic lesions and which could not be distinguished from Wernicke's encephalopathy. In three of the 19 cases there was a total necrosis within the mamillary bodies. By re-examining the mamillary bodies of 12 known alcoholics without any evidence for an ischemic impact we could affirm that total necrosis may fit into the spectrum of Wernicke's encephalopathy. Our findings demonstrate that the morphological changes in the mamillary bodies due to thiamine deficiency and those due to hypoxia-ischemia may be identical.