Purpose: In this prospective study, we compared the long-term cognitive and behavioral effects of lamotrigine (LTG) and carbamazepine (CBZ) in patients with newly diagnosed or untreated partial epilepsy. Methods: This was a multicenter, open-label, randomized study that compared monotherapy with LTG and CBZ in newly diagnosed or untreated patients with partial epilepsy. We employed an 8-week titration period and a 40-week maintenance period. Neuropsychological tests, Symptom Check List-90, and QOLIE-31 were assessed at baseline, 16 weeks, and 48 weeks after drug treatment. A group-by-time interaction was the primary outcome measure and was analyzed by use of the linear mixed model. Results: A total of 110 patients were eligible and 73 completed the 48-week study (LTG, n = 39; CBZ, n = 34). Among the cognitive tests, significant group-by-time interaction was identified only in phonemic fluency of Controlled Oral Word Association Task (p = 0.0032) and Stroop Color–Word Interference (p = 0.0283), with a significant better performance for LTG group. All other neuropsychological tests included did not show significant group-by-time interactions. Among the subscales of Symptom Check List-90, significant group-by-time interactions were identified in Obsessive-Compulsive (p = 0.0005), Paranoid Ideation (p = 0.0454), Global Severity Index (p = 0.0194), and Positive Symptom Total (p = 0.0197), with a significant improvement for CBZ group. QOLIE-31 did not show significant group-by-time interactions. Conclusion: Our data suggest that epilepsy patients on LTG have better performance on phonemic fluency and the task of Stroop Color–Word Interference than do patients on CBZ, whereas patients on CBZ had more favorable behavioral effects on two subscales and two global scores of Symptom Check List-90 than did patients on LTG.