Weeds are both a harmful crop pest and an important component of biodiversity. Moreover, herbicide use must be reduced to limit its impact on environment, and weed control must now combine numerous management techniques with partial efficiency aiming at preventing weed occurrence. Biological regulations of weeds by other biotic components could also contribute to control infestations. Thus, new cropping systems are needed, combining numerous techniques and aiming at both maximising weed-related biodiversity and minimising weed harmfulness.[br/] Weed dynamics models are increasingly used to design innovative cropping systems but usually only consider weed densities and crop yield. The objective of the present work was to illustrate with an existing model how these models can be improved to (1) integrate new knowledge and management techniques, (2) integrate biotic interactions and to assess weed-related harmfulness and biodiversity, and then (3) used to evaluate and design innovative cropping systems.[br/] The FLORSYS model is to date the only multispecific weed dynamics model that integrates the effect of all cropping system components (crop succession, all management techniques) in interaction with pedoclimate. It is a mechanistic (i.e. process-based) model which synthesizes data from different experiments and teams and easily evolves to integrate new knowledge or management techniques. For instance, additional processes (e.g. the blocking of weed seed rain to soil seed banks by permanent grass canopies) were recently added to FLORSYS to adapt it to temporary grassland in arable crop rotations (Doisy et al., this conference). These complex models are though more difficult to validate with field observations, a step still underway for FLORSYS.[br/] To integrate biotic interactions with other organism in weed models, new submodels are needed to quantify the effect (1) of cropping systems on the new organism (here the fungus [i]Gaeumannomyces graminis[/i] var. [i]tritici[/i] responsible for the take-all disease in cereals), (2) of the new organism on weeds (here the decrease in seed production of diseased weed plants) and (3) of weeds on the new organism (here pathogen propagation by diseased weed plants). To assess weed-related harmfulness and biodiversity in cropping systems, the predicted weed densities must be translated into indicators. In the present work, five harmfulness indicators (crop yield loss, technical harvest problems, harvest pollution, field infestation and additional crop disease incidence caused by fungi-transmitting weed species) and five biodiversity indicators (species richness, species equitability, seed resource for birds and insects, pollen/nectar resource for pollinators) were constructed and connected to FLORSYS (Mézière et al., this conference). These models are interesting for a large range of applications. At short-term, they can be used to optimize individual management techniques in different weed flora contexts, e.g. FLORSYS was used to evaluate different crop sowing strategies (sowing densities and patterns, crop associations etc.). More interestingly, these models can simulate existing cropping systems over several years and with different climate scenarios to test their long-term and climatic robustness. FLORSYS was thus used to evaluate a large range of cropping systems identified in farm surveys and to identify cultural practices pertinent for controlling weed harmful and preserving biodiversity. However, the major interest of models is to test prospective cultural techniques and cropping systems. FLORSYS is now used to evaluate the changes in agricultural practices (e.g. simplified tillage and rotations, no-till, temporary crops, Colbach et al., this conference). Currently, our team is working on a simulation-based methodology to design prospective cropping systems with low herbicide use, minimising weed-related harmfulness and maximising biodiversity.