OBJECTIVE Positron emission tomography with 2-deoxy fluoroglucose positron emission tomography (18-FDG-PET) is widely used in the pre-surgical evaluation of subjects with epilepsy, but little is known of its usefulness in a non-surgical population. METHODS We analyzed the sensitivity of PET as a diagnostic tool in a large unselected population of epilepsy subjects. Pre-surgical and non-surgical portions of this population were individually assessed as well. The relationship of PET abnormalities to other neurodiagnostic tests was examined. Statistical assessment relied primarily on contingency tables (chi-square tests), with ANOVA or non-parametric assessment used as necessary. RESULTS While PET was more likely to identify areas of decreased metabolism in the surgical population than in the non-surgical populations, it nevertheless found a significant number of abnormalities in the total population and in the non-surgical group alone. Even in groups in which the clinical diagnosis was unknown, abnormalities were found 40% of the time. PET was useful as an exclusionary diagnostic tool for non-epileptic seizures (NES) and primary generalized epilepsies (PGE) with sensitivity, specificity, and accuracy > 90%. The PET was somewhat more sensitive than magnetic resonance imaging (MRI) in finding abnormalities in the total population, but was less sensitive than electroencephalography (EEG). CONCLUSIONS PET may be a useful diagnostic tool in the general epilepsy population even when a definitive clinical diagnosis is not suggested by other modalities.