We examined the ability of preoperative memory performance to distingu
ish between patients who had been diagnosed as having left (LTLE, n =
31), right (RTLE, n = 37), and extra-(ETLE, n = 17) temporal lobe foca
l epilepsy. All patients eventually underwent surgical resections. Ana
lyses indicated that the ETLE group performed better than the RTLE gro
up on nonverbal memory measures and better than the LTLE group on verb
al memory measures. Discriminant function analyses indicated that use
of a combination of measures that assess different aspects of memory w
ere of significant value in distinguishing between patients with focal
TLE and ETLE. This approach, as compared the use of single measures,
improved classification rates of all three groups. The best single pre
dictor of group membership, an index of verbal learning, yielded a 47%
overall correct classification rate, with sensitivities ranging from
25 to 59%, and performed at worse than chance levels in classifying RT
LE patients. A multivariate approach, which included an index of verba
l and nonverbal learning, incidental nonverbal memory, and consolidati
on of organized and rote verbal material, yielded a 65% correct classi
fication rate, with sensitivities ranging from 57 to 75%. This compare
s favorably with other noninvasive techniques for lateralizing epilept
ogenic lesions.