Offshore steel structures are exposed to harsh environmental conditions. At the same time, they are susceptible to fatigue due to cyclic loads from wind, waves, and operation as well as traffic. Both the steel structures and the corrosion protection have a limited lifetime. When the corrosion protection reaches the end of its service life, the steel is exposed to free corrosion. To continue the operation, the influence of corrosion on the fatigue behaviour has to be evaluated. Therefore, new methods need to be developed to accurately assess the actual condition of corroded components and integrate this information into (remaining) service life predictions.Motivated by these challenges, this work began with investigations on corroded steel plates (also referred to as base material). Initially, methods for capturing and numerically modelling the actual corroded condition were developed. A corroded steel surface was converted into a numerical model using a 3D scan, allowing the calculation of stress concentrations caused by corrosion. Subsequently, fatigue tests were conducted on pre-corroded steel samples. Prior to fatigue tests, the samples were 3D scanned to characterize the state of corrosion and determine the stress concentrations based on the real surface. It was demonstrated that the hotspots of stress concentrations correlated with the crack locations observed in the fatigue tests. To incorporate the stress concentrations into the fatigue analysis according to local fatigue concepts, various methods for considering the micro-support effect were investigated. The application of the implicit gradient model (IGM) yielded the most reliable results with the lowest scatter in the derived stress-life (SN) curve based on notch stresses. Using the parameters derived for the IGM and the corresponding notch stress SN-curve, a quantitative relationship between the real surface of a corroded steel component and the fatigue life was established.Steel structures often include numerous welded connections, which behave differently to corrosion exposure, due to existing stress concentrations from the welding geometry and residual stresses from the welding process. Hence, additional fatigue tests were conducted on corroded fillet and butt-welded specimens. In addition, 3D scans and residual stress measurements were performed. Subsequently, the fatigue tests were evaluated using local fatigue approaches, taking the real weld geometry from the 3D scans and the measured residual stresses into account. It was shown that stress concentrations had a significant impact on the fatigue life prediction. Additionally, it was demonstrated that residual stresses played a crucial role and must be considered for assessing the remaining fatigue life of corroded steel components. The application of local methods minimized the scatter in the SN-curves and thus increased the reliability. It was shown, that the consideration of the actual condition of corroded components in the calculation of fatigue strength is possible, also for welded samples.Finally, the replica technique was introduced. With the replica technique, imprints of corroded components can be created during inspections of the structures. These imprints are then scanned in the laboratory and analysed using the methods presented in this work to determine their stress concentrations, which are correlated with the endurable load cycles. The replica technique can be applied to both welded and non-welded components. This allows the transfer of the findings from executed fatigue tests and numerical calculations to real structures.