The control of the erosive streaming in an agricultural catchment requires good knowledge of agricultural practices impacts on hydrological processes. These impacts have to be evaluated both in time and space. Our work aims at proposing a method to assess the impact of agricultural practices on runoff emission.. The method is based on the collection of the present agricultural practices -crop acreage, crop rotations and tillage operations- by interviewing all the farmers of a watershed. These data are used to run two models: (1) a model (DIARE) which simulates for each farmland hectare a runoff curve according to a chosen climatic scenario on a day by day basis (2) a model (STREAM) which simulates the runoff concentration network and computes the runoff volumes of the considered catchment for a specific rainfall event. Using the case study of the Bourville catchment (Haute-Normandie, France) we show that the two models allows to establish a comprehensive assessment of runoff production at catchment and farm level. Firstly, DIARE allows to identify over the year the high runoff producing periods. Secondly, STREAM allows to point out fields of the catchment which contribute to runoff at a given date. Both results can be analysed for a given farm or for the whole catchment. We plan to use these two models simultaneously to generate a range of technical improvements that can be then discussed with farmers. This method could really improve dialogue between farmers and advisers insofar as it enable farmers to see the consequences of changes they could implement on their farm, both at the catchment level (with a focus on space) and at their farm level (with a focus on time).