In this study the influence of the accuracy reached within fiber orientation distributions calculated with Moldflow and Moldex3D is examined. The models implemented in Moldflow and Moldex3D are based on the same publications of Tucker et al. These new approaches consist of model parameters that can be optimized dependent on the used material. In a first step an optimization loop is established that uses experimentally gathered orientation data and simulated ones. In order to realize this optimization loop, first samples of the same material with varying thicknesses are injection molded. Secondly 3D measurements on these plates at various positions of the fiber orientation tensor elements are performed based on µ-CT image stacks. In parallel, the production of the plates is simulated in Moldflow and Moldex3D and the tensor elements are taken on the same positions as in the experiments. In the optimization loop the model parameters are now modified as long as the difference between the curves gathered by experiment and in simulation has reached a minimum. In a second step a real component is simulated using the optimized parameters and for comparison the default values. Applying a reality relevant load case on these components the difference between the two models based on the simulated fiber orientations is shown. The material model used is based on a micro mechanical approach using a reverse engineering step to optimize the deformation behavior of this material. The material used is a technical, short glass fiber reinforced thermoplastic, widely used in automotive industries.