In 15 patients we hypothesized the origin of epilepsies to be 'extratemporal' based on videotaped seizures and surface EEG. Neuropsychological tests and neuroimaging (CAT, MRI, and PET scans) were then compared to the hypothesized ictal sites. Neuropsychological tests were abnormal in 86.6% and FDG-PET scans were abnormal in 73%. The neuropsychological tests and PET localized or lateralized areas of dysfunction to the same sites as electroclinical characteristics did in 85% of patients (P less than 0.01, Fisher exact test). No statistically significant correlation between lesion sites on CT and MRI and the ictal origin was observed due to the high proportion of normal or non-specific scans. These observations should be verified in a larger series of extratemporal seizures.