Optimal function of left ventricular assist devices (LVADs) depends on proper alignment of the inflow cannula (IC). Quantitative guidelines for IC angulation are lacking because of variation in cardiac geometry and difficulty in analyzing three-dimensional (3D) cannula orientation relative to the left ventricle (LV). Based on contrast-enhanced computed tomography images from five normal and five clinically malpositioned IC cases in patients with HeartMate II LVADs, we developed a method for 3D quantification of IC malpositioning. Using Mimics image software (Materialise, Leuven, Belgium), the native heart, major arteries, and LVAD were segmented to create patient-specific 3D models, allowing LV cavity volume and long-axis length to be measured directly. The deviation of the IC was quantified in a cylindrical coordinate system at the IC insertion point relative to the mitral valve and septum, and IC occlusion was assessed by the distance between cannula inlet and the proximal endocardium. Compared with normal cases, patients with malpositioned pumps had shorter LV length (p = 0.03) and reduced pump pocket depth (p = 0.009). Malpositioned pumps may experience greater obstruction by the nearby myocardium. This quantitative 3D modeling tool may help identify different modes of pump malalignment and migration and may facilitate preoperative planning and minimally invasive approaches via virtual LVAD implantation.