Corrosion can significantly affect the fatigue strength of steel structures. On the one hand, it can cause stress concentrations by changing the surface geometry, and on the other hand, it can lead to embrittlement of the material. In welded steel components, sharp notches are already present due to the weld geometry, which may interact with the notches from corrosion. In addition, residual stresses exist in the fatigue-relevant area, which can also be partially modified by surface preparation methods such as clean blasting. In the current standards, all the mentioned effects are considered by reducing the design stress-life (SN) curves, in which all aforementioned effects are superimposed. One possibility to consider the effects separately is the application of local concepts, with which the influence of the geometry, here obtained from 3D-scans, the influence of measured residual stresses, and the influence of material can be considered individually. For butt and fillet welds investigated in this study, it could be shown that by considering the real geometry via a reverse engineered 3D-solid model, the locations of crack initiation in the fatigue tests, measured by digital image correlation (DIC), matched with the numerically predicted locations in most cases. Moreover, it was shown that by taking into account the real geometry and residual stresses, the use of local approaches can result in a substantially reduction of the scatter in SN-curves from Tσ = 2.17 to Tσ = 1.26, demonstrating the great reliability of the applied method.