Non-invasive risk stratification for ventricular fibrillation (VF) in Brugada syndrome (BrS) has not been fully evaluated. The aim of this study was to assess the utility of signal-averaged Holter electrocardiogram (Holter SAECG) and 12-lead Holter electrocardiogram (Holter ECG) after a pilsicainide provocation test for non-invasive risk stratification in BrS. We enrolled 30 consecutive patients with BrS [divided into 2 groups: the VF group, those with a previous history of VF (n = 10); and the non-VF group, those without a history of VF (n = 20)] and 10 control subjects without type 1 ECG. We evaluated late potentials [LP: filtered QRS (f-QRS), RMS40, and LAS40] on the Holter SAECG for 4 h after the pilsicainide provocation and in the same patients on another day without performing the pilsicainide provocation. Furthermore, we measured QRS duration and QTc interval in leads V2 and V5, and J amplitude in lead V2 on the Holter ECG after the pilsicainide provocation. On the Holter SAECG, the f-QRS at 1 h and LAS40 at 3 h after the pilsicainide provocation were significantly larger in the VF group than in the non-VF group (f-QRS at 1 h: 113.9 ± 8.9 vs. 104.9 ± 8 ms; p = 0.01, LAS40 at 3 h: 45.4 ± 5.9 vs. 35.5 ± 7.4 ms; p < 0.001). The receiver-operating characteristic curve analysis for a single parameter of VF occurrence was determined [f-QRS at 1 h: area under the curve (AUC) 0.8, with sensitivity 80% and specificity 80%; and LAS40 at 3 h: AUC 0.87, with sensitivity 90% and specificity 75%]. On the Holter ECG, there were no significant differences in these parameters between the VF and non-VF groups. In conclusion, the LP after the pilsicainide provocation using Holter SAECG may be useful for risk stratification of VF episodes in patients with BrS.