In this study we examined 37 subjects with a diagnosis of intractable frontal lobe epilepsy (FLE) based on non-invasive pre-surgical evaluation. Twenty-six underwent chronic intracranial ictal recordings (CIR) with video monitoring; 20 of these went on to surgical resection. Eleven underwent surgery without CIR. Retrospectively, we determined that 19 had pure FLE, 12 had frontal plus extrafrontal epileptogenic zones, and six others did not have FLE. We analysed the whole group and individual categories to evaluate the determinants of surgical outcome. Sixty percent of the pure frontal group is seizure free with all having > or = 75% reduction. The frontal-plus group had only 10% seizure free with 70% having > or = 75% reduction. Being in the pure frontal group was associated with better outcomes than the 'frontal-plus' group (P < 0.05; chi-square). Subjects with FSIQ > or = 85, focal pathologies and 18FDG-PET scans which were normal or had focal abnormalities (P < or = 0.05, all, chi-square) were more likely to have excellent outcomes. MRI abnormalities, surface EEG, and location and size of resection were not predictive of surgical outcomes. Rasmussen's encephalitis, incomplete surgical strategies and bilateral foci were apparent in those with poor outcomes, and surgical size predicted post-operative deficits (chi-square; P < 0.001). We conclude that careful, hypothesis-driven implants and operating procedures can result in good surgical outcomes for frontal lobe epilepsy subjects even when lesions are not apparent on routine neuroimaging.