Soil-plant N dynamics lie at the heart of some of the questions being asked of researchers by farmers, environmentalists and policy makers. Our aim in this paper is to highlight areas in which research is needed to address these questions. Although we have a general understanding of many processes, fundamental understanding of the processes of the soil-plant system is not complete. Improved understanding of crop and soil processes should lead to the continuous improvement of simulation models, which are able to integrate the complex effects of management and environmental factors. However, farmers cannot wait for the achievement of perfect models, but need researchers to put their current knowledge to use. We suggest that for both crops and soils, diagnostic measurements be used in conjunction with the best of, or a combination of, current models. Development work should be carried out with all possible speed to draw suitable models and diagnostics together. We stress the importance of conducting research to understand and improve N efficiency at a range of scales and indicate the need for the involvement of related disciplines, such as statistics, to allow the development of robust guidelines and methodologies for up- and down-scaling measurements and models. Above all, it is essential that our understanding of the processes of soil-plant dynamics continues to underpin the development of strategies for dynamic optimisation and improve simulation models that are used in fertiliser recommendation systems.