This retrospective description of a surgical series is aimed at reporting on indications, methodology, results on seizures, outcome predictors and complications from a 20-year stereoelectroencephalography (SEEG) activity performed at a single epilepsy surgery centre. Prospectively collected data from a consecutive series of 742 SEEG procedures carried out on 713 patients were reviewed and described. Long-term seizure outcome of SEEG-guided resections was defined as a binomial variable: absence (ILAE classes 1-2) or recurrence (ILAE classes 3-6) of disabling seizures. Predictors of seizure outcome were analysed by preliminary uni/bivariate analyses followed by multivariate logistic regression. Furthermore, results on seizures of these subjects were compared with those obtained in 1128 patients operated on after only non-invasive evaluation. Survival analyses were also carried out, limited to patients with a minimum follow-up of 10 years. Resective surgery has been indicated for 570 patients (79.9%). Two-hundred and seventy-nine of 470 patients operated on (59.4%) were free of disabling seizures at least 2 years after resective surgery. Negative magnetic resonance and post-surgical lesion remnant were significant risk factors for seizure recurrence, while type II focal cortical dysplasia, balloon cells, glioneuronal tumours, hippocampal sclerosis, older age at epilepsy onset and periventricular nodular heterotopy were significantly associated with seizure freedom. Twenty-five of 153 patients who underwent radio-frequency thermal coagulation (16.3%) were optimal responders. Thirteen of 742 (1.8%) procedures were complicated by unexpected events, including three (0.4%) major complications and one fatality (0.1%). In conclusion, SEEG is a safe and efficient methodology for invasive definition of the epileptogenic zone in the most challenging patients. Despite the progressive increase of MRI-negative cases, the proportion of seizure-free patients did not decrease throughout the years.