Patient-specific simulations are widely used to investigate the local hemodynamics within realistic morphologies. However, pre-processing and mesh generation are time consuming, operator dependent, and the quality of the resulting mesh is often suboptimal. Therefore, a semi-automatic methodology for patient-specific reconstruction and structured meshing of a left coronary tree from biplane angiography is presented. Seven hexahedral grids have been generated with the new method (50,000-3,200,000 cells) and compared to nine unstructured tetrahedral grids with prismatic boundary layer (150,000-3,100,000 cells). Steady-state blood flow simulation using Computational Fluid Dynamics (CFD) has been used to calculate the Wall Shear Stress (WSS). Our results (99 percentile, area-weighted and local WSS values along a line) demonstrate that hexahedral meshes with respect to tetrahedral/prismatic meshes converge better, and for the same accuracy of the result, six times less cells and 14 times less computational time are required. Hexahedral meshes are superior to tetrahedral/prismatic meshes and should be preferred for the calculation of the WSS.