The intracarotid amobarbital test (IAT) examines hemispheric memory and language. We set out to determine whether memory performance on the IAT correlated with seizure relief after anterior temporal lobectomy in 117 patients with refractory epilepsy. The IAT assessed recognition memory performance for nine items with correction for false-positive recognitions. We then compared performance of one hemisphere with that of the other, defining a correctly lateralized memory deficit as worse performance when using the hemisphere containing the operated temporal lobe than when using the other hemisphere. The analysis included concurrent factors that might also affect outcome, such as age at first risk for epilepsy, presence or absence of tumor, and Full Scale IQ. A discriminant function analysis demonstrated that patients with a correctly lateralized memory deficit on the IAT had an increased probability of being seizure-free following surgery after controlling for other predictors. The performance of the nonoperated temporal lobe related to outcome, although less strongly. The magnitude of the difference in performance between the two hemispheres and the performance of the operated hemisphere did not relate to outcome. Patients who became seizure-free had an earlier age at first risk than did those with persistent seizures, and tumor presence weakly correlated with postoperative outcome. IQ did not correlate with outcome. We conclude that the IAT predicts seizure relief after anterior temporal lobectomy independent of other known risk factors we examined.