Background and purpose: To assess the efficacy of various antiepileptic drugs (AEDs) for controlling post-stroke epilepsy. Methods: This nationwide cohort study was conducted by using data from 2004 to 2008 on new occurrence of post-stroke epilepsy obtained from the National Health Insurance Research Database of Taiwan. The examined AEDs were phenytoin (PHT), valproic acid (VPA), carbamazepine (CBZ) and new AEDs. Recurrent seizures requiring either emergency room (ER) visits or hospitalization were used to measure the efficacy of seizure control. The Kaplan-Meier failure curve and Cox proportional hazard regression analyses were used to compare the risk of seizure recurrence in patients taking various AEDs. Results: In all, 3622 late-onset post-stroke epilepsy patients were selected. Overall, 1.05 and 0.70 recurrent seizure incidences occurred per 100 person-months based on ER visits [95% confidence interval (CI) 0.95-1.15] and hospitalizations (95% CI 0.62-0.78), respectively. The incidences of ER visits for patients using different AEDs were 1.26, 0.70, 0.43 and 0.38 per 100 person-months for PHT, VPA, CBZ and new AEDs, respectively. Compared with patients using PHT, the adjusted hazard ratios for ER visits were 0.56 (95% CI 0.42-0.74; P < 0.001), 0.37 (95% CI 0.18-0.75; P = 0.006) and 0.28 (95% CI 0.15-0.52; P < 0.001) for patients using VPA, CBZ and new AEDs, respectively. The adjusted hazard ratios of hospitalizations for seizure recurrence yielded similar results. Conclusions: This large nationwide, population-based study demonstrated that late-onset post-stroke epilepsy patients using VPA and new AEDs have better seizure control than those using PHT as demonstrated by lower risks of ER visits and hospitalization.