Cover crop (CC) mixtures appear as a promising agroecological tool to suppress weeds and maintain crop productivity while reducing chemical inputs such as herbicides and nitrogen fertilizers. The weed suppressive effect of CC and CC nitrogen release have been related to CC composition, soil resource availability and CC termination methods but the relative effect of these factors and their interactions remain poorly documented, especially in the case of CC mixtures. Most studies have focused on the subsequent spring crop, and hence, lacked to investigate long-term carry over effects of CC. A two field: year experiment was conducted to investigate the effect of CC mixture (2 or 8 species including or not legume species, compared to a bare soil control), resource availability (water and nitrogen) at CC sowing and CC termination method (rolling and glyphosate, compared to a winterkill control) on weed biomass and productivity of two subsequent unweeded, unfertilized and direct seeded crops (spring barley and winter linseed in year 1 and 2, respectively). In the context of reduction of herbicide use, our experiments showed that the use of CC to reduce weed biomass in the subsequent crop was not relevant when CC were terminated by winter or rolling since CC probably protected weeds from winter instead of creating a physical barrier for weed emergence, weed community composition in crop being mainly represented by weed volunteers. Weed biomass was the main driver of the subsequent crop productivity but the latter was enhanced after fertilised or legume-based CC, probably because of a higher nitrogen release. No relevant long-term effect of CC on the subsequent crop was observed in our experiment, probably because nitrogen was used by the previous crop and weed community composition changed by year. Our results highlight the importance CC composition and management for weed suppression and crop productivity in low-intensity biodiversity-based cropping systems.