We studied patients with documented temporal lobe seizures to evaluate the predictive value of computed tomography (CT), magnetic resonance imaging (MRI), and F-18 fluorodeoxyglucose positron emission tomography (FDG-PET) for surgical therapy and the relationships between these tests and the pathologic diagnoses. CT detected abnormalities in 32.5%, with an accuracy of 19% when accuracy was defined as congruence with electrophysiologic studies. MRI detected abnormalities in 81%, with an accuracy of 67%. FDG-PET detected abnormalities in 85%, with an accuracy of 82%. Pathologic change was detected in 79% of the excised temporal neocortex, 65% of amygdalae, and 93% of hippocampi. After follow-up periods of 20-71 months (mean 41 months), 67% of patients were free of seizures and 94% had at least a 90% reduction in seizure frequency. There was no relationship between the type of abnormality on MRI or the type of pathology and postoperative outcome. Better outcomes were associated with focal or regional ictal onsets as recorded by surface EEG. Worse outcomes were associated with hypometabolism that extended outside the temporal lobe. Pathologic change in the temporal neocortex was associated with extension of hypometabolism outside the temporal lobe.