Long fiber-reinforced thermoplastics have favorable mechanical properties, low manufacturing costs and superior lightweight characteristics. However, the configuration of the reinforcing fibers changes significantly throughout the entire production process, reflected in mechanisms such as fiber attrition, fiber alignment, and fiber matrix separation. The complexity of the process-microstructure-property relationship limits the use of this material class in a wider range of lightweight applications. This work presents a contribution to gain a better understanding of the underlying physics of fiber motion during molding and to obtain a theoretical link between processing and the fiber microstructure. As part of this work, novel characterization techniques were developed to overcome the shortcomings of conventional measurement approaches. The measurement concepts comprise methodologies to characterize the fiber orientation, fiber concentration and fiber length by applying sophisticated techniques, which include combining image processing with micro computed-tomography and optical measurement systems. Applying the developed measurement techniques, a range of experimental studies were conducted to investigate the process-induced fiber microstructure. Plate geometries were injection molded at varying nominal fiber concentrations to investigate fiber matrix separation, fiber alignment, and fiber breakage. The experiments revealed substantial variation in the fiber configuration and correlation between the microstructural properties. The substantial fiber matrix separation and fiber breakage found in this work refute the common assumption of uniform fiber concentration and fiber length in molded parts. Additionally, a new experimental setup based on a Couette rheometer was developed to study fiber length reduction under highly controlled conditions, which isolated the impact of processing conditions on fiber attrition. Finally, the generated experimental data are applied to evaluate models predicting the process-microstructure relationship. All models fail to provide acceptable results and the application of these models as truly predictive tools is limited. It was shown that a holistic approach is necessary to capture the process-induced change in fiber configuration, which necessarily includes the interdependencies of the microstructural properties.