SUMMARYObjective: The objective of the study was to characterize interictal 18-fluorodeoxyglucose-positron emission tomography ( 18 FDG-PET) whole-brain voxel-based metabolic patterns among distinct subtypes of temporal lobe epilepsy (TLE), as defined by stereo-electroencephalography (SEEG) and to determine predictive value of PET result on postoperative outcome. Methods: Fifty-four consecutive patients with pharmacoresistant TLE were enrolled retrospectively after a comprehensive presurgical evaluation. This evaluation defined: 7 lateral TLE, 17 mesial TLE, 14 "plus" TLE, and 16 bilateral TLE. Whole-brain voxelbased brain metabolism was studied in each group of patients, in comparison to 23 healthy subjects, and individual classification was evaluated by cross-validation using the found clusters. An 18 FDG-PET index was moreover calculated for each patient, based on the individual Z-score of the most significant cluster extracted on the comparison between patients' subgroup and healthy subjects. Logistic regression analysis was used to estimate factors associated with postoperative outcome (Engel's classes III-IV vs. I-II), including age, gender, disease duration, seizure frequency, as well as magnetic resonance imaging (MRI) and PET findings. Results: Different patterns of hypometabolism were found inside and outside the epileptogenic zone, among patients with distinct subgroups of TLE, in comparison to healthy subjects (p < 0.001, corrected for the cluster). At individual level, cross-validation showed satisfactory discrimination between the four groups with 71.4-88.2% overall accuracy. Multivariate analysis shows that 18 FDG-PET index was the only significant predictor of postoperative outcome to distinguish between Engel's classes I-II and III-IV (p = 0.037). Significance: Overall, this whole-brain voxel-based analysis validates specific patterns of hypometabolism, inside and outside the EZ, in distinct subgroups of patients with TLE, as defined by SEEG gold standard, and in relation with postoperative outcome.