Finding efficient and environmentally sustainable methods for arable weed management is a current challenge in agroecosystems. Weed species disperse over scales larger than the field. Landscape-scale management could, therefore, reinforce field-scale strategies for innovative cropping systems. In particular, the introduction of semi-natural habitats, such as permanent meadows, which are unsuitable habitats for annual arable weeds, could potentially reduce weed densities in crops. We developed a spatially explicit, landscape-scale model of annual arable weed population dynamics including the main components of weed management at cropping system level: crop rotation, tillage regime, intensity of herbicide use and mechanical weeding. The variability of responses to these factors for different weed species functional types was considered by modelling four functional response groups. We used this model to investigate the effects of increasing the proportion of permanent meadows in the landscape and of the spatial arrangement of meadows on weed density in crops. With this model, increasing the proportion of permanent meadows significantly decreased mean seed density of weeds in crops. Furthermore, weed density in crops was found to be lower for landscapes in which the meadows were scattered, rather than clustered together in large zones. However, these spatial and landscape-scale effects were much weaker than the local effect of cropping systems. Weed density in permanent meadows was more sensitive to the proportion of meadows and to their spatial arrangement than weed density in crops, although the absolute value of weed density remained low in this unfavourable habitat.