Quantitative data on the potential losses of crop yield due to weeds is limited. Approaches to quantify weed damages considered only individual weed species, and have not included the compounding of effects at the community level. Moreover they did not integrate the influence of agricultural practices and environmental variables on the weed-crop relationship. Here we analysed a big data sets to study relations between crop management, weed communities’ indicators and crop yield and then develop path analysis models to disentangle the confounding effects of cropping system on weeds and crop production. Flora surveys (N= 651) were performed from 2006 to 2013 on 150 fields in the Fénay study site (Dijon, eastern France). Farmers were interviewed to know their cropping systems (N=564) and recorded their yield (N=564). The main crops were winter wheat (195 fields, 227 flora surveys), winter barley (94, 110), oilseed rape (92, 103) spring barley (53, 63) and mustard (45, 56). Data were also collected on the 11 others crops but on less fields (<25 fields, <27 flora surveys for each crop). Pearson correlations were performed between quantitative variables (farming practices, weed flora indicators) and yields.