A ‘system management’ approach of biological weed control in crops is proposed and compared with other method's of biological weed control. It is based on the management of a weed pathosystem in order to maximize the natural spread and disease severity of a native or naturalized pathogen. This approach may be well-suited to situations where it is necessary to control single weed species in crops, and where no immediate and complete control is required, the production of large amounts of the agent is rather limiting (e.g. when using biotrophic fungi), and/or the importation of an exotic agent is not possible. This strategy provides fundamental knowledge of underlying mechanisms of crop production systems and is aligned with the view of modern agro-ecology, in which complete eradication of weeds is not desirable. The fundamental research required for a successful application of the ‘system management’ approach will be illustrated with the biological control project of Senecio vulgaris L. using the naturalized rust fungus Puccinia lagenophorae Cooke. A five-step procedure, together with selected results, will be presented. Main emphasis is given to the infection window, the study of the genetic structure of the plant and pathogen population, and the management of the infection conditions (a) to maximize the spread of the disease and the impact on the plants, and (b) to minimize the development of resistant plant populations. Joint application of herbicides at low doses, additional necrotrophic pathogens, and of biochemicals interfering with the weed's defence also will be envisaged, as well as their integration into general pest control practices. In this regard, biological weed control agents have to be seen as stress factors, not as weedkillers, and biological weed control as an integral part of a well-designed pest management strategy, not as a sole cure.