Several alternative engineering models are available for the use in analysis of offshore wind turbines. However, it is not always clear which of the models will yield the most accurate or sufficiently conservative results. This paper investigates the effect of using two alternative soil-structure interaction models and two wind coherence models. The focus is on assessing how these modelling choices influence the predicted long-term fatigue damage in the support structure. The two soil models are a macro-element model and a p-y-curve model with Rayleigh damping. This gives differences in both the damping and stiffness properties of the turbine model. The differences between the two soil models tend to decrease as the turbine size increases. The wind coherence models considered are the Kaimal spectrum with exponential coherence and the Mann uniform shear turbulence model. The Kaimal model predicts the highest response at low frequencies, while the Mann model gives the highest response predictions at higher frequencies. Which turbulence model predicts the highest long-term fatigue damage is then determined by the natural frequencies, rotor and blade passing frequencies of the different turbines.