Cardiac ventricular arrhythmias can cause sudden death. Despite known genomic contributions, multigenic risk predictors are limited. The genetics of arrhythmias and cardiomyopathies overlap, with additional overlap with epilepsy. To improve genetic risk prediction, we assemble a cohort with non-ischemic ventricular arrhythmias and controls lacking cardiac diagnoses. Here, we integrate 18 polygenic scores; variants from clinical gene panels for coding regions of cardiomyopathy, arrhythmia, and epilepsy genes; and noncoding regulatory regions mapping to those genes. Polygenic scores alone hold prognostic value. Rare coding variants identify cumulative risk extending beyond known pathogenic/likely pathogenic variants. We also find enrichment of ultrarare regulatory variation. A risk predictor that combines all variant classes outperforms any single class or subset and replicates in a validation cohort. This combined genomic arrhythmia propensity score (GAPS) identifies high-risk individuals even among those who lack known primary pathogenic variants. This integrated approach serves as a model for other complex traits.