Question: Which weed traits respond to the community assembly filters imposed by cropping regimes and how do they respond? Which crop characteristics (traits or aspects of field management) represent the strongest filters on weed traits? Location: France. Methods: In 561 arable fields, we measured associations between crop characteristics and (i) mean values of quantitative weed traits at each site and (ii) composition of optimally defined weed functional types, based on both qualitative and quantitative weed traits and typologies. The crop characteristics included crop height, propagule size, family and sowing date; we also used basic types (winter wheat, maize, etc.) for comparison. Results: Crop sowing date was strongly related to many weed traits, whereas crop type was not the strongest predictor of any trait. With later sowing of crops, weeds started flowering later, germinated later and had shorter flowering periods. Sowing date was also associated with the distribution of Raunkiaer life forms, while crop architecture was associated with the distribution of weed heights and a C-S-R classification. Conclusions: Cropping regimes can usefully be summarized by the crop sowing date. Phenological traits of weeds as well as classifications based on life form and C-S-R are important descriptors of weed community composition. Such findings may help predict the effects of particular crop rotations, novel crop varieties and invasive weed species.