The study of fatigue crack growth is an important issue to assess the fatigue strength of engineering components and service life. The aim of this master thesis is a numerical, analytical and experimental analysis of crack propagation in welded joints. A comparative study by the aid of the software packages Franc2D and nCode is performed for different structural details. Special focus is laid on a standard single edge crack plate, and welded joints ranging from non-load carrying to partial penetrated load-carrying transversal stiffeners. This enables a direct comparison to recently published results by the IIW-community. Thereby, the assessment based on Franc2D is slightly conservative because of the user-dependent internal integration procedure settings. In case of the tool nCode, an accurate analytical calculation of the stress intensity factor (SIF) is necessary and the results are in good accordance to the Round robin IIW-results. The experimental work focuses on testing and microscopic evaluation of block constant amplitude fatigue tests with and without overloads. The investigated material ranges from common construction steel S355 to high-strength steel S690. Single overloads cause macroscopic beachmarks which are detectable by LOM, but striation markers within the polycrystalline microstructure are only recognisable by SEM. For the investigated specimens characteristic striation markers are evaluated and basically match the numerically obtained crack propagation results. Finally, an extensive set of numerical Franc2D calculations are performed using numerous different material parameters given by state-of-the-art guidelines, recommendations and published literature. Thereby, it is shown that the initial crack length is the most important factor for crack propagation and total life time. A detailed study of the initial crack length shows that a value of 0.01mm is well applicable for the fatigue-tested high strength steel specimens. A comparison between simulation and experimental results proves that all crack growth parameters lead to a safe design if the initial crack length is not lower than this value. Summarized, for a defined minimum crack length, all evaluated material parameters exhibit a fatigue life which is below the recommended scatter range of welded structures and therefore the investigated methods are well applicable for a conservative fracture mechanical assessment.