PURPOSE: We investigated whether visual and quantitative ictal EEG analysis could predict surgical outcome after anteromesial temporal lobectomy (AMTL) in which mesial structures, basal, and temporal tip cortex were resected. METHODS: We retrospectively reviewed 282 presurgical scalp-recorded ictal EEGs (21- to 27-channel) from 75 patients who underwent AMTL. We examined the pattern of seizure onset (frequency, distribution, and evolution) and estimated the principal underlying cerebral generators by using a multiple fixed dipole model that decomposes temporal lobe activity into four sublobar sources (Focus 1.1). We correlated findings with a 2-year postoperative outcome. RESULTS: Sixteen patients had seizures with a well-lateralized, regular 5 to 9-Hz rhythm at onset, that most often had a temporal or subtemporal distribution. All patients became seizure free after surgery. In 51 patients, seizure onset was remarkable for lateralized slow rhythms (<5 Hz), which sometimes appeared as periodic discharges, were often irregular and stable only for short periods (<5 s), and had a widespread lateral temporal distribution. Among these a favorable surgical outcome was encountered in patients with seizures having prominent anterior-tip sources ( 16 of 17 seizure free), whereas those with dominant lateral or oblique sources had a less favorable outcome (three of 14 and 13 of 18, respectively). Irregular, nonlateralized slowing characterized seizure onsets in eight patients. Three patients became seizure free after surgery. CONCLUSIONS: Both visual and quantitative sublobar source analysis of scalp ictal EEG can predict surgical outcome in most cases after AMTL and complement non-invasive presurgical evaluation.