By using [11C]flumazenil-positron emission tomography ([11C]FMZ-PET), we have previously shown that reductions of central benzodiazepine receptors (cBZRs) are restricted to the hippocampus in mesial temporal lobe epilepsy (mTLE) caused by unilateral hippocampal sclerosis (HS). Receptor autoradiographic studies on resected hippocampal specimens from the same patients demonstrated loss of cBZRs that was over and above loss of neurons in the CA1 subregion. Here, we report the first direct comparison of in vivo cBZR binding with [11C]FMZ-PET and ex vivo binding using [3H]FMZ autoradiography. We applied a magnetic resonance imaging-based method for partial volume effect correction to the PET images of [11C]FMZ volume of distribution ([11C]FMZ Vd) obtained in 10 patients with refractory mTLE due to unilateral, histologically verified HS. Saturation autoradiography was performed on the hippocampal specimens obtained from the same patients, allowing calculation of receptor availability ([3H]FMZ Bmax). After correction for partial volume effect, [11C]FMZ Vd in the body of the epileptogenic hippocampus was reduced by a mean of 42.1% compared with normal controls. [3H]FMZ Bmax, determined autoradiographically from the same hippocampal tissue, was reduced by a mean of 42.7% compared with control hippocampi. Absolute in vivo and ex vivo measurements of cBZR binding for the body of the hippocampus were significantly correlated in each individual. Our study demonstrates that reduction of available cBZR on remaining neurons in HS can be reliably detected in vivo by using [11C]FMZ-PET after correction for partial volume effect.