We report the results of a protocol for choosing candidates for temporal lobectomy using a standard battery of objective tests without intracranial electrodes. We assigned each test a level of importance, and an algorithm was used to determine whether temporal lobectomy could be performed. Fifty-one patients (total pool, 103 patients) met protocol requirements and had an anterior temporal lobectomy with a mean follow-up of 39.4 months (range, 21 to 64 months), most remaining on anticonvulsant therapy. Eighty percent are seizure free, 12% have <3 seizures per year or only nocturnal seizures, and 8% have >80% reduction in seizure frequency. One-third of patients who failed protocol criteria did not have temporal lobe seizures when studied with intracranial electrodes. We analyzed and modified the algorithm after comparing these patients with others who were poor candidates for temporal lobectomy. We conclude that this protocol is effective and recommend using such an objective algorithm.