Background: Arrhythmic risk stratification in patients with Brugada syndrome remains controversial. Several recent reports have highlighted the possible role of late potential (LP) on the signal-averaged electrocardiogram (SAECG) as a noninvasive risk stratification tool in Brugada syndrome, but further prospective study is required before its general applicability. Objective: The purpose of this study was to investigate the role of LP in arrhythmic risk stratification of Brugada syndrome patients. Methods: Forty-three patients with Brugada syndrome were enrolled and divided into a symptomatic group (group A, n = 24) and an asymptomatic group (group B, n = 19). SAECG was performed to analyze the characteristics of LP in all subjects. The prospective study was conducted to observe the occurrence of arrhythmic events. Results: LP was positive in 22 (91.7%) of 24 patients in the symptomatic group and in 7 (36.8%) of 19 patients in the asymptomatic group. During mean follow-up of 33.8 ± 9.0 months, the incidence rate of arrhythmic events was 72.4% (21/29) in LP-positive [LP(+)] patients compared with 14.3% (2/14) in LP-negative [LP(-)] patients. Multivariate Cox proportional hazard analyses revealed that the presence of LP had the most significant hazard ratio of 10.9 (95% confidence interval 1.1-104.3, P = .038), with sensitivity of 95.7%, specificity 65.0%, positive predictive value 75.9%, negative predictive value 92.9%, and predictive accuracy 81.4%. Kaplan-Meier curves plotted for event-free survival according to LP showed a significant difference between LP(+) and LP(-) patients (log rank, P = .003). Conclusion: The results of this study support the role of LP detected by SAECG in arrhythmic risk stratification of Brugada syndrome patients. © 2009 Heart Rhythm Society.