Cardiac magnetic resonance (CMR) imaging has emerged as a pivotal tool in evaluating and managing arrhythmia risk across a spectrum of cardiac pathologies, including ischemic heart disease (IHD) and various cardiomyopathies. CMR provides several imaging features, such as the extent and distribution of ventricular scar tissue and atrial fibrosis, as valuable imaging features for guiding clinical decision-making and improving prognosis prediction in patients with cardiac arrhythmias. Recent advancements in Artificial Intelligence (AI) have introduced novel models that leverage CMR data for precise risk stratification and prediction of arrhythmic events. Among the most clinically significant arrhythmias, atrial fibrillation (AF) and ventricular arrhythmias (VAs) have garnered particular attention due to their substantial impact on morbidity and mortality. This chapter explores the pros and cons of different CMR modalities—including late gadolinium enhancement (LGE), T1 and T2 mapping, and four-dimensional (4D) flow imaging—in assessing arrhythmia risk in the context of diverse cardiac diseases. Furthermore, it provides an overview of the progress in AI-driven CMR applications, highlighting their potential to enhance arrhythmia risk assessment and contribute to personalized patient care.