The objective of this study was to assess which features of temporal lobe proton magnetic resonance spectroscopic imaging (1H-MRSI) are associated with satisfactory surgical outcome in patients with intractable temporal lobe epilepsy and bilateral hippocampal atrophy. We studied 21 patients with intractable temporal lobe epilepsy and bilateral hippocampal atrophy defined by magnetic resonance imaging volumetric measurements who underwent surgical treatment. 1H-MRSI was used to determine the relative resonance intensity ratio of the neuronal marker N-acetylaspartate to creatine + phosphocreatine (NAA/Cr) for mid and posterior temporal lobe regions of the left and right hemisphere, as well as an asymmetry index. Values lower than 2 SDs below the normal mean were considered abnormal. We used Engel's classification to assess surgical outcome with respect to seizure control. Eleven patients (52%) were in class I-II and 10 (48%) were in class III-IV. All 21 were operated on the side of maximal electroencephalographic (EEG) lateralization. Concordant lateralization of decreases in NAA/Cr to the side of surgery and normal NAA/Cr values in the contralateral posterior-temporal region were significantly associated with good surgical outcome: 11 (69%) of 16 patients with 1H-MRSI abnormalities concordant with EEG lateralization and none of the 5 patients with nonconcordant 1H-MRSI had a good outcome (class I-II); 10 (77%) of 13 patients with normal NAA/Cr contralateral to the EEG lateralization versus 1 (12.5%) of 8 of those with NAA/Cr reduction contralateral to EEG lateralization were in class I-II. Regression correlation analysis showed significant linear correlation between the midtemporal NAA/Cr relative asymmetry ratio and surgical outcome; the greater the asymmetry, the better the outcome. We conclude that discriminant 1H-MRSI features associated with favorable surgical outcome in patients with temporal lobe epilepsy and bilateral hippocampal atrophy were (1) concordant 1H-MRSI lateralization, (2) a greater side-to-side asymmetry of NAA/Cr, and (3) an absence of contralateral posterior NAA/Cr reduction.