In the context of precision nitrogen (N) management, since 2002, the Walloon Region (Belgium) transposed the European Nitrate Directive 91/676/EEC under the Sustainable Nitrogen Management for Scheme Agriculture (PGDA). An ensemble of "good practices" were defined to reduce the environmental pressure linked to excessive N management while maintaining productivity of Belgian farmers. However, using process-based crop growth models, substantial progress could again be achieved to increase the N-use efficiency.

 Crop models are powerful tools to assess the impacts of different cropping systems inputs, i.e. agro-environmental conditions, management practices and weather, on the crop harvestable organs. On an other hand, the complexity linked to the N management lies in the fact that the expected impact of a given practice is delayed in time from its application. Quantifying the risk associated to representative climatic time-series is a suitable approach to cope with this uncertainty. Therefore, to improve the decision-making process, a complete methodology was designed to simulate yield distributions as an answer to the most probable upcoming climatic conditions interacting with multiple N management strategies.

 The STICS soil-crop model was used to simulate the growth of a winter wheat crop. A 30-years historical weather database was analysed using the LARS-Weather Generator. A set of 300 synthetic weather time-series representative of the climatic conditions in the area were then derived. Decision rules, integrating agronomic, economic and environmental objectives were then derived in order to determine the optimal N practices.

 First results showed that, 3 years out of 4, substantial gains could be made on the N fractions applied usually at flag-leaf stage. Following the economic constraints (grain selling price and N cost) and the application of environmental taxes, 10 to 30kgN.ha^-1 could be systematically saved, increasing the N use efficiency by 5 to 20%, in comparison to the current Belgian Farmers' practice.