We retrospectively compared ictal technetium 99m hexamethylpropyleneamineoxime single-photon emission computed tomography (SPECT) and interictal 18F-fluorodeoxyglucose positron emission tomography (PET) in 35 patients with well-lateralized temporal lobe epilepsy (TLE). Based on SPECT scans the two observers correctly lateralized seizure foci with certainty in 89% of patients; interobserver agreement was excellent. Both observers incorrectly lateralized the seizure focus on two SPECT scans; one error was explained by rapid electroencephalographic spread to the contralateral side and for the other patient, isotope was injected during a brief aura. Based on PET scans, observers correctly lateralized the foci with certainty in 63% and with lesser confidence in 83%; four incorrect lateralizations were made by one observer and none by the other. PET interobserver disagreement was explained by differences between observers in weighting the relative hypometabolism in medial and lateral temporal regions. The detection rate for PET was lower in the absence of structural imaging abnormalities (60 vs 87%). PET yielded correct lateralizations in the 2 patients for whom SPECT interpretation was difficult. We conclude that both ictal SPECT and interictal PET are sensitive methods for the lateralization of TLE, but SPECT can be interpreted with greater certainty and is more sensitive when magnetic resonance imaging findings are negative. False lateralization is rare with ictal SPECT and can be explained when interpreted in conjunction with electroclinical data. Both investigations have complementary roles when localization is difficult.