Early repolarization pattern (ERP) has recently been associated with idiopathic ventricular fibrillation and with cardiovascular mortality in the general population. We aimed to identify electrocardiographic tools to differentiate the “malignant” form of ERP from benign ERP in a population-based study. We retrospectively assessed the prevalence of ERP by recording electrocardiograms in 1,161 southwestern French subjects 35 to 64 years old. ERP was defined by an elevation of the J point ≥1 mm in 2 consecutive leads excluding leads V₁ through V₃. We categorized ERP as notching or slurring pattern as located in inferior and/or lateral leads and measured the J-point elevation amplitude. ST segment after ERP was categorized as ascendant or horizontal/nonascendant and T waves as negative or positive. Association of ERP with all-cause and cardiovascular mortalities was assessed by adjusted Cox proportional hazard models. ERP was found in 157 subjects (13.3%). During a mean follow-up of 14.2 ± 2 years, 77 subjects died (6.6%), of whom 24 (2.1%) died from cardiovascular causes. Subjects with ERP had an increased hazard ratios for all-cause mortality (2.45, 95% confidence interval [CI] 1.44 to 4.15, p = 0.001) and cardiovascular mortality (5.60, 95% CI 2.27 to 11.8, p = 0.001). The highest risk was found for notching ERP and ERP with a nonascendant/horizontal ST segment, yielding when associated increased hazard ratios of 3.84 (95% CI 2.14 to 6.92, p = 0.001) and 8.75 (95% CI 3.48 to 22.0, p = 0.001) for all-cause and cardiovascular mortalities, respectively. Conversely, a slurring ERP or ascendant ST segment was not associated with increased mortality. ERP localization, J-point elevation amplitude, or T-wave morphology did not distinguish benign from malignant forms of ERP. In conclusion, ERP with notching pattern and horizontal/descendant ST segments was associated with the highest risk of all-cause and cardiovascular deaths. These electrocardiographic patterns may be used for risk stratification in subjects with ERP.