The loads imposed on, amongst others, offshore structures can vary considerably with time. Lifetime prediction methodologies need to consider possible acceleration and retardation of the crack growth rate due to load sequences. Models based on a linear accumulation of damage will have a limited accuracy and are not considered as a sufficiently valuable asset in lifetime prediction of structures subjected to variable amplitude loading. This necessitates more complex nonlinear damage evolution models that can be applied in a so-called cycle-by-cycle analysis. In this paper, a comparison is made between three cumulative damage models (Miner, modified Miner and weighted average) and two yield zone models (Wheeler and Willenborg). Experimental data of fatigue crack growth in offshore steel specimens subjected to sequential loading is used as basis of the comparison. The modified Miner model is the most promising of the cumulative damage models but the determination of the parameter α requires laboratory tests. Evaluation of the effects of variation in the model input parameters on estimated lifetime reveals a large influence for the Miner and weighted average approaches.