A non-invasive presurgical protocol for temporal lobe epilepsy (TLE) based on ‘anatomo-electro-clinical correlations’ was tested. All consecutive patients with suspected TLE and seizure history > 2 years were entered in theprotocol which included video-EEG monitoring and MRI. Three different TLE sub-syndromes (mesial, lateral, mesiolateral) were identified by means of a combination of anatomical, electrical and clinical criteria. A tailored operation for each sub-syndrome was offered. Patients with seizure history < 2 years, MRI evidence of temporal mass lesion and concordant interictal EEG and clinical data by-passed video-EEG monitoring and were directly scheduled for surgery. Lesionectomy was performed without video-EEG recording in 11 tumoral TLE patients. Of 146 patients studied with video-EEG, 133 received a diagnosis of TLE. Four of them were excluded for neuropsychological risks, 8 refused surgery, and 121 were operated. Of 132 consecutive patients who underwent surgery, 101 had at least 1 year of follow-up. They were divided in an ‘hippocampal sclerosis/cryptogenic’ group (n = 57) and a ‘tumours/cortical organization disorders group’ (n = 44). In the first group, extensive temporal lobectomy (ETL) was performed in 40 patients, antero-mesial temporal lobectomy (AMTL) in 17 patients. At follow-up, 47 (82.5%) patients were seizure-free. In the second group, lesionectomy plus ETL was performed in 23 patients, lesionectomy plus AMTL in 6 patients, and lesionectomy alone in 15 patients. Thirty-nine (88.6%) patients were seizure-free. Our findings suggest that different TLE sub-syndromes can be accurately identified using non-invasive anatomo-electro-clinical data and can be effectively and safely treated with a ‘tailored’ operation.