The present major agricultural issues are to feed the world and reduce negative environmental impacts. To this end, organic farming appears as a promising solution. However organic farming has several drawbacks such as difficult weed management. Indeed weeds can reduce crop yields. Therefore there is a need for improved decision support tools for weed management in organic farming. An existing weed competition model actually predicts the effect of early multispecies weed density, both on organic wheat yield loss and on the weed density at flowering stage. However main existing models do not take into account the activity of end-users, e.g. farmers, during model design. Therefore we analysed weed information acquisition by farmers using the dynamic environment theory to design a decision support system that takes into account end-users. We interviewed eight French organic farmers. We analysed interview data using a coding scheme inspired by dynamic environment theory. Our results show that weed quantity was the information most frequently collected by organic farmers both for short- and long-term crop management. This information was compatible with early weed density, the main input of the previously developed models. Findings also show that procedures for gathering information and processing depended on farmer profiles. We also show that a conceptual model based on dynamic situations and a coding structure were appropriate for taking into account the information elaborated by end-users. Finally we propose further design of a decision support system for tactical organic weed management using a participatory approach.