Quantitative assessment of mitigation measures for nitrogen (N) pollution requires adequate models, good knowledge of catchment functioning and a thorough understanding of agricultural systems and stakeholder constraints. The current paper analyses a set of results from simulations, with two models, of agricultural changes in two catchments in different contexts with different constraints. The results show that reducing N inputs and increasing grassland areas are the most efficient measures, not only because they reduce N fluxes in streams but also because they enhance N use by agriculture and the whole catchment system. Introducing catch crops, hedgerows and riparian buffers are interesting complementary measures but of limited impact when implemented alone. These results are sensitive to the way mitigation measures are translated into model inputs, and their operational implications are discussed.