The GRAFS approach can be used to describe nitrogen flows in agri-food systems. It is based on an empirical relationship which establishes a link between crop rotation-scale nitrogen yields and total nitrogen inputs following a hyperbolic shape, with a single parameter Ymax representing the theoretical maximum yield. Here, we aim to clarify the role of agro-pedo-climatic variables as drivers of the spatial variability of Ymax. This study lays the groundwork to enhance the predictive capacity of the GRAFS model within the context of climate change, changing agricultural practices and land use.