Various simulation models uniformly project a decline of the cereal and silage maize area as well as ruminant production in the EU-15 in the course of decoupling of direct payments. In contrast, model results are heterogeneous with respect to the direction of the decoupling effect on oilseed and pasture as well as voluntary set aside area. The model type (behavioural partial or general equilibrium model, or programming model) is not found to have a systematic effect on model results. It is rather the ad hoc assumptions about the effectiveness of direct payments on production which differ widely and drive model results to a large extent.