Noninvasive imaging of cardiac electrical activity is of great importance and can facilitate basic cardiovascular research and clinical diagnosis and management of various malignant cardiac arrhythmias. This dissertation research is aimed to investigate a novel physical-model-based 3-dimensional cardiac electrical imaging (3DCEI) approach. The 3DCEI approach is developed by mathematically combining high-density body surface electrocardiograms (ECGs) with the anatomical information. Computer simulation study and animal experiments were conducted to rigorously evaluate the performance of 3DCEI. The simulation results demonstrate that 3DCEI can localize the origin of activation and image the activation sequence throughout the three-dimensional ventricular myocardium. The performance of 3DCEI was also experimentally and rigorously evaluated through well-controlled animal validation studies in both the small animal model (rabbit) and large animal model (canine), with the aid of simultaneous intramural recordings from intra-cardiac mapping using plunge-needle electrodes inserted in the ventricular myocardium. The clinical relevance of 3DCEI was further demonstrated by investigating 3DCEI in cardiac arrhythmias from animal models with experimentally-induced cardiovascular diseases. The consistent agreement between the non-invasively imaged activation sequences and its directly measured counterparts in both the rabbit heart and canine heart implies that 3DCEI is feasible in reconstructing the spatial patterns of ventricular activation sequences, localizing the arrhythmogenic foci, and imaging dynamically changing arrhythmia on a beat-to-beat basis. The promising results presented in this dissertation study suggest that this cardiac electrical imaging approach may provide an important alternative for non-invasively imaging cardiac electrical activity throughout ventricular myocardium and may potentially become an important tool to facilitate clinical diagnosis and treatments of malignant ventricular arrhythmias.