Multimodality evoked potential (MEP) data from over 300 comatose head-injured patients suggest that central nervous system dysfunction of the brain stem and/or hemispheres can be localized with this noninvasive neuroelectric technique. Based on this work, decerebrate motor posturing and prolonged coma are not associated with brain-stem dysfunction but rather with dysfunction of the hemispheres, while absent pupillary and oculocephalic responses are correlated with brain-stem dysfunction alone. However, the accuracy with which MEP data localized human brain-stem or hemispheric dysfunction could not be confirmed by pathological correlation because of low mortality and the small number of autopsies obtained in the patients who died. Therefore, this study was undertaken in an animal model of brain-stem lesion. Complete brain-stem transections were made at the cervicomedullary junction, the medulla just caudal to the eighth nerve, and at the intercollicular region. All cortical visual evoked potential (VEP) peaks were reduced in amplitude and delayed by each of the brain stem transections, but none of the peaks was abolished. In spite of brain-stem transection, VEP's can be used to gain information about hemispheric function. Somatosensory (SEP) and auditory cortically generated evoked potentials (AEP) were abolished by these brain-stem transections, but early-latency brain-stem SEP and AEP data could accurately localize specific areas of brain-stem dysfunction caused by the lesions. Observations made on human MEP data seen to be confirmed by these animal experiments. Correlations between human and cat MEP data are discussed.