Patients with epilepsy under consideration for temporal lobectomy frequently undergo language mapping through direct cortical stimulation to avoid post-operative deficits. More recently, it has become possible to map language using the less invasive technique of Positron Emission Tomography (PET). These methods differ fundamentally in that the former maps language by disrupting brain function, the latter by activating these functions. There has not previously been a direct comparison of these diverse methods of language mapping. In this study, 7 epilepsy patients about to undergo temporal lobectomy for seizure control were administered identical language tasks during H2 15 0 PET and electrical stimulation mapping (ESM). Images from PET and CT following electrode implantation were placed in register with each subject's MR scan to allow the direct comparison of the mapping techniques with respect to anatomical structures. The results showed a high degree of correspondence between PET-activated regions and areas of language disruption from ESM. PET also showed areas of significant activation which were apparently involved in, but seemingly not critical for, language performance.