Injection moulding is a discontinuous process that makes it possible to produce highly reproducible plastic moulded parts in series at low cost. Due to fluctuations in a wide range of influencing variables, shrinkage and warpage as well as the associated dimensional accuracy of a plastic moulded part are very difficult to predict. The aim of this thesis was to evaluate the predictability of the dimensional variation of injection moulded parts using injection moulding simulation. A wide range of influencing variables in the areas of the process, the material, the environment, the mould, the measurement technology and the machine were considered. In particular, the fluctuation range of the influencing variables is to be shown and their influence on the dimensional accuracy investigated. In this work, almost 60 potentially dimensionally relevant influencing variables were analysed. In addition to extensive literature research, discussions with experts and individual investigations, screening tests were also carried out to determine the significant influencing variables and their range of fluctuation. Using the Monte Carlo method, all significant influencing variables were considered on the basis of their probability of occurrence and fluctuation range and integrated into the injection moulding simulation as realistically as possible. The Autodesk Moldflow® Insight 2023 (AMI) injection moulding simulation software (Autodesk Inc., USA) was used for the calculations, while the warpage geometries determined with it were evaluated using the GOM Inspect 2020 metrological analysis software (Carl Zeiss GOM Metrology GmbH, Germany). The simulation result was compared with the measurements of the selected article during 14 months of series production. In addition, the manufacturability of the required mould part tolerances was evaluated with the help of DIN ISO 20457. The predicted dimensional scatter showed similarities with the series measurement values over shorter periods of time, whereby the shrinkage was slightly overestimated in the simulation. In practice, however, jumps in measured values occurred over the entire period, which could not be reproduced. The comparison with DIN ISO 20457 showed that the series measurement values harmonised well with the tolerance, i.e. all measured data was within the tolerance of the respective test characteristics. Based on the simulations, Minitab (Minitab GmbH, USA) and STASA QC (STASA Steinbeis Angewandte Systemanalyse GmbH, Germany) were used to create a linear and nonlinear process model for the mathematical description of the dimensional scatter, which showed a high correlation to the simulation results and to each other during validation. Finally, various improvement potentials were identified with regard to influencing variables and injection moulding simulation in order to increase the prediction accuracy of dimensional scattering.