Due to the increasing strength of lightweight design, glass fiber reinforced polymers are used because of the excellent specific material properties. In order to enable safe operation of the components, material characterizations for dimensioning must be carried out under static, quasi-static and cyclic loads. The comparisons of the simulated and the fiber orientation distribution determined by µCT measurements made within this work show that the deviations that occur correspond to the current state of the art and that discrepancies of 29% are achieved in the critical area. The single-stage W¨ohler tests carried out in the present work at R = 0.1 and R = -1 aim to determine the influence of the microstructure of longitudinal and transverse specimens from 15 different sampling positions of the standardized, injected molded CAMPUS plate to the local fatigue strength. The evaluations of the W¨ohler tests show that the longitudinal samples (L- specimens) have significantly longer service life values than the transverse samples. In addition, the L-samples show the trend that the fatigue strenght increases significantly from the side near the gate to the end of the flow path and with an increase in the distance to the center of the plate. The transverse specimens behave in the opposite way here. When the mean stress is increased, both the longitudinal and the transverse samples show reductions of around 40% in the tolerable stress amplitude at one million cycles. Through the use of mini test specimens and the additional comparison with short test samples, an attempt is made to determine the statistical size influence. In context of these investigations a statistical size influence cannot be determined. For this, further experiments should be carried out. In addition, a continuous simulation path starting from shell filling simulation up to service life calculation in FEMFAT ® could be shown in the present work, which enables an alternative for insufficient 3D filling simulations. The computer-aided service life estimation shows usable (6% deviation in the tolerable stress amplitude at one million load changes) for the longitudinal samples and increases (16%) results for the transverse samples, despite the various simplifications. For more accurate evaluations it will be necessary to improve the material models as part of further investigations.