In temporal lobe epilepsy (TLE) patients without lesions, major hippocampal sclerosis, or atrophy on magnetic resonance imaging (MRI), the localizing power of [11C]flumazenil (FMZ) and 2-[18F]fluoro-2-deoxy-D-glucose (FDG) was compared using high-resolution positron emission tomography (PET) studies and individually coregistered MRI scans. Following complete clinical, neuropsychological, and electrophysiological evaluation, benzodiazepine receptor density was assessed using the FMZ equilibrium method. Thirty minutes later, interictal FDG-PET was performed under resting conditions. PET images were matched to three-dimensionally coregistered, T1-weighted MRI. Each temporal lobe (TL) was divided into 12 volumes of interest. The regional FMZ data were normalized with respect to average cortical values. For each patient the right-left asymmetries of rCMRGlc and normalized FMZ data were calculated. In 7 to 10 patients, mesial TL structures showed reduced FMZ binding, with a decrease by at least 10% in the affected TL. Reductions of 10% or more of rCMRGlc usually were more widespread than FMZ reductions and often involved lateral temporal cortex. The regions of most pronounced disturbances are not necessarily identical in both methods. Three patients had a complex correspondence of lateralization with PET, neuropsychological, and EEG data. In 4 patients, lateralization was less clear from EEG or neuropsychological results but was still consistent with lateralization by PET. In 3 of 10 patients, however, major discrepancies were found. These data suggest that the combination of neuropsychological testing, EEG, and MRI-guided FMZ- and FDG-PET will help to select patients with clearly defined epileptogenic foci especially in mesial TLE. Even in cases without MRI lesions, TL epileptic foci can be lateralized with consistency across the methods; FMZ-PET shows the pathologic focus more circumscribed than FDG-PET.