As the current agricultural practices in the Hauts-de-France region result in soil erosion, nitrate leaching and a decline in biodiversity, agroforestry systems (AFs) may be an alternative to conciliate productivity with lower environmental impact (Dupraz and Liagre, 2008). We set up the first AF experimental site to study its agro-economic and environmental performance in local agro-pedoclimatic conditions. The experimental site was established in autumn 2018 on an 18-ha plot in Ramecourt on a deep luvic cambisol with a silt loam texture developed on a flint clay. Due to a high silt content (73% silt) and a low organic matter content (2%), as well as a slope of 8%, the plot is highly affected by channel erosion. Modalities with or without nitrogen-fixing trees in AF treatment are compared with sole-crop (CC) and pure-forest control (FC) plots (Figure 1A) according to a randomised block design with 3 replicates. Tall trees in rows are intercalated with 9 species of shrubs (Figure 1B) and will be intercropped in AFs by sugar beet, potato, wheat, barley and flax. The tree density is 50 and 430 trees ha-1 for AFs and FC respectively, and the average size of the microplots is 0.9 ha. Using this experimental approach, we hypothesised that AFs should limit soil erosion, restore soil fertility and biodiversity, improve natural-resource use efficiency and water quality, reduce inputs and increase farmers’ incomes.