Sunflower is submitted to several major pathogens. Modeling is a key tool to help design. Integrated Pest Management strategies to control them. A new qualitative modelling approach is currently under progress using the IPSIM platforrn (Aubertot and Robin, 2013) for sunflower. It aims at predicting injury profiles on sunflower as a function of cropping practices, soi!, weather, and the surroundings of the considered field. Based on a literature review and expert knowledge, hierarchical deterrninistic bayesian networks were developed. Independent datasets were used to assess their quality of prediction. This communication will present: i) a first draft of IPSIMSunflower; ii) the evaluation of its predictive quality; iii) examples of simulation to help design IPM strategies to contrai the disease; iv) a discussion on the limits and benefits of the approach, along with perspectives.