Take-all of wheat is a major disease in most wheat production areas throughout the world. To date, there exists no single method of control that can alone provide high levels of efficacy.Some techniques, such as delaying the sowing date, reducing sowing density, using a fungicide seed treatment or applying an ammonium form of nitrogen fertilisation have all shown partial control. By combining some of the most promising of these treatments, growers can receive advice for take-all management for use mainly at a local scale. However, there is currently no reliable method to help farmers optimise their choice of cropping systems in order to improve their control of take-all. Based on a simple epidemiological model that takes into account primary infection, from particulate inoculum in the soil, and secondary infection, as disease spreads from plant to plant, we conducted, during the last ten years, epidemiologic research to examine (i) the effect of wheat crop management practices on take-all dynamics,(ii) the relationship between these dynamics and yield losses. Using a data base of 204 site/year/cropping system combinations that vary according to their take-all management, we finally propose a set of models that predict take-all incidence and its impact on yield performance, as functions of soil characteristics, climate and agricultural practices.