Over a 10-year period 100 temporal lobe operations were performed for non-lesional temporal lobe epilepsy (TLE). Seventy-two patients had conventional anterior temporal lobectomy (ATL) and 28 had selective amygdalohippocampectomy (AH). Forty-three ATL patients (60%) became seizure free compared to 6 AH patients (21%) (P < 0.005). An unsatisfactory outcome was obtained in 11 (15%) ATL patients and 14 (50%) AH patients (P < 0.01). Fourteen AH patients subsequently had a second ATL procedure; 6 improved, 3 becoming seizure free. There were no differences between the two groups in the incidence of mesial temporal sclerosis (MTS) or normal pathology. However, ATL patients had a higher concordance rate of preoperative investigations which may have made them more favourable surgical candidates. It was concluded that ATL was more likely to produce a seizure free outcome than AH and should be used as the routine surgical procedure for temporal lobectomy. However, AH can give a satisfactory outcome and may be preferred when a patient has good verbal memory and all preoperative tests are concordant.