Computational fluid dynamics (CFD) methods based on in vivo three-dimensional vessel reconstructions have recently been shown to provide prognostically relevant hemodynamic data. However, the geometry reconstruction and the assessment of clinically relevant hemodynamic parameters may depend on the used imaging modality. This study compares geometric reconstruction and calculated wall shear stress (WSS) values based on magnetic resonance imaging (MRI) and computed tomography (CT). Both imaging methods were applied to a same 2.5-fold upscale silicon model of the left coronary artery (LCA) main bifurcation. The original model is an optically digitized post mortem vessel cast. This digitized geometry is considered as a "gold standard" or original geometry for the MRI versus CT comparative study. The use of the upscale model allowed generating a high resolution CT raw data set with voxel size of 0.156 x 0.156 x 0.36 mm(3) and a high resolution MRI data set with an equivalent voxel size of 0.196 x 0.196 x 0.196 mm(3) for corresponding in vivo conditions. MRI based reconstruction achieved a mean Hausdorff surface distance of 0.1 mm to the original geometry. This is 2.5 times better than CT based reconstruction with mean Hausdorff surface distance of 0.252 mm. A comparison of the calculated mean WSS shows good correlation (r = 0.97) and good agreement among the three modalities with a WSS of 0.65 Pa in the original model, of 0.68 Pa in the CT based model and of 0.67 Pa in the MRI based model.