Nitrate fluxes caused by intensive agriculture are responsible for pollution of fresh water including drinking water. Controlling this diffuse pollution requires the application of action programs on drinking water catchment areas (WCA) to support farmers in the management of cropping systems reducing nitrogen losses. This involves the use of tools to evaluate the effect of these systems on nitrate leaching. Models for simulation of water and nitrogen luxes are theoretically powerful tools to link the combinations of practices within cropping systems and the nitrate content of groundwater recharge. By quantifying nitrogen losses at plot and WCA levels, the outputs from these models should represent very relevant assessment indicators. However, it is accepted tha these models are hardly used by the actors in charge of the implementation of these actions on the WCA. The objective of this study has been to contribute to the improvement of the knowledge of these models as environmental assessment tools. We therefore aimed to evaluate their ability to be operational tools by specifying their usefulness and their usability for non-scientific actors. A critical analysis of the tools currently used has shown that the local stakeholders almost exclusively use simple predictive indicators such as the Burns model or measured indicators such as a residual soil mineral nitrogen. The STICS crop model and the AGRIFLUX environmental model were tested on a multi-year scale and at three experimental sites in France where nitrogen losses were documented by lysimeters or porous cups. These models were also tested on a small WCA located in northeastern France. All these tests were conducted without calibration of the parameters under conditions of use voluntarily close to those of local actors. The performances abtained with AGRIFLUX are better because of its greater precision in this context of use, linked in part to the non-consideration of plant growth and the use of yields as on input variable to calculate nitrogen uptake. On the other and, drainages simulated by AGRIFLUX and STICS are often very close and the formalism adopted by STICS facilities the parametrization compared to that of AGRIFLUX. It appears that the choice of an environmental model such as AGRIFLUX, which requires fewer parameters, seems easier for actors. A crop model such as STICS would be very difficult to use for actors. A crop model such as STICS would be very difficult to use for non-specialists because of the optimization of plant parameters, which is essential for accurately assessing nitrogen fluxes. Since these tests have also confirmed the major role of soils in nitrogen losses, it seems advisable to target the simulations on the soils most vulnerable to nitrate losses. Finally, there are interesting prospects for increasing he use of biophysical models, particularly by working on specific formalizations of output variables in order to achieve the qualities of simplicity, speed, reliability and efficiency sought by non-specialist users.