This study aimed to determine the reliability of clinical history and seizure semiology for distinguishing between frontal lobe seizures (FLS) and temporal lobe seizures (TLS). FLS patients (n=23) were consecutively identified through an epilepsy surgery database. TLS patients (n=27) were selected randomly from 238 patients who had undergone temporal lobe surgery for epilepsy. The criterion standard for seizure localization was the location of resective epilepsy surgery that controlled seizures for a minimum of 2 years. Blinded comparisons of 13 historical information items (HII) and 19 video-recorded semiologic features (VSF) were made. We identified 3 HII (sex, history of febrile convulsions, and history of generalized tonic-clonic seizures) and 2 VSF (fencing posturing and postictal confusion) that significantly distinguished between FLS and TLS. The multivariate analysis model correctly identified 87% of FLS patients and 74% of TLS patients. No single HII or VSF is sufficient for distinguishing between FLS and TLS. A model integrating multiple HII and VSF may assist in this differentiation, but some patients still may be misclassified.