The geometry of vascular system is an important determinant of blood flow in health and disease. There is a strong geometric component to atherosclerosis in coronary heart disease since lesions are preferentially located at bifurcation points and regions of high curvature. The influence of these local structures on recirculation and deleterious shear stresses and their role in plaque development is widely accepted. Over time, researchers have turned to MR, CT, or biplane images of vascular trees to faithfully capture these features in the flow simulations. Historically, this has taken the form of labor-intensive manual reconstructions from morphometric measurements based on the centerline, whereby small idealized subsets of vascular trees are developed into computational grids. With improved imaging, image processing, and geometric reconstruction algorithms, researchers have begun to develop geometrically accurate computational models directly from the medical images. This chapter provides an overview of contemporary methods for image processing, centerline detection, boundary condition definition, and grid generation of both clinical and research images of cardiovascular structures.