Sixty patients with temporal lobe epilepsy were classified into reading deficient (RD; n = 21) and non-reading deficient (non-RD; n = 39) groups. Selective deficits in verbal or nonverbal memory, consistent with side of seizure onset, were evident in the non-RD patients. Both verbal and nonverbal memory performance were reduced equivalently in individuals with RD, regardless of side of seizure onset. As a result, memory tests that were accurate in identifying side of seizure onset in the non-RD group were not as accurate in the RD group. When individual cases were classified using a clinically applicable decision rule, significantly more RD patients were either unclassifiable or incorrectly classified than were non-RD patients. Findings suggest that preoperative memory data obtained from individuals with epilepsy and evidence of RD may not be as valid an indicator of side of seizure onset as are those obtained from patients without RD.