This diploma thesis deals with a shape optimization to improve the fatigue behavior of a cylinder head. To achieve this goal a sequentially coupled thermal-stress analysis is carried out using the FEA software Abaqus/Standard. The resulting data on mechanical stresses is decomposed in average stress and a fluctuation around this mean using the software tool FEMFAT, in order to carry out a fatigue analysis. Based on the results from this analysis a shape optimization is performed using the software tool TOSCA Structure. In the next step, this optimization process is automized using Python scripts. For this purpose a plug-in for Abaqus/CAE is created which manages all settings that are necessary for the entire optimization process. When executed, the plug-in generates all required input files for Abaqus/Standard, FEMFAT and TOSCA Structure automatically. In the first part of this work, the fundamentals of fatigue and structural optimization, in general, and of shape optimization, in particular, are reviewed. Additionally, the calculation process of the software tools used plus the optimization workflow are described. Secondly and following several preliminary tests, this method is applied to investigate the potential of a further weight reduction of a cylinder head as well as the automation of the optimization workflow itself. It is found that both a weight reduction of the cylinder head and an increase in high cycle fatigue safety are possible. Finally, it can also be concluded that the Abaqus plug-in developed for this work allows the postprocessing of the entire fatigue and optimization results within Abaqus/CAE, besides the preprocessing for FEMFAT and TOSCA Structure.