AbstractIt is a common practise to predict fatigue crack propagation rates for specimens containing residual stresses using the weight function or finite element method. Due to combined applied load and internal residual stresses, the stress intensity factor Ktot at the crack tip is calculated and used to predict the resulting fatigue crack propagation rates in conjunction with empirical crack growth laws. The calculation of Ktot normally implies pure linear elastic behaviour and the validity of the superposition law. For cracks growing through areas of high compressive residual stresses and subsequent transition areas from compressive to tensile residual stresses, this assumption is not necessarily valid. In this case the definition of non-linear contact conditions on the crack faces becomes necessary to describe the problem in a physically sound way. The presented study discusses the resulting differences in the prediction results for the case of an aluminium C(T)100 specimen containing a residual stress pattern typically found after welding processes.