A framework of dynamic useroriented interactive forecast system is proposed. The newgeneration forecast system contains an initial userend module for configuring the dynamic forecast target, the physical predictive, downscaling components, userend professional models, and a useroriented assessing module in association with decisionmaking at the userend. A case study involving application of current global ensemble forecasts (TIGGE) of rainfall for hydrological users in Linyi, a region rich in rivers and reservoirs in eastern China, is conducted in order to highlight interactive involvement of the userend information during forecast process. An iteratively selfimproving forecast system, involving changing decisionmaking information throughout a flood season, a changing threshold for floodleading rainfall depending on previous weather and concurrent hydrological conditions, and a dynamic useroriented assessing process of concurrent skill and uncertainty of the prediction, is built up. It is demonstrated how a forecast of floodleading rainfall leads to a better warning signal for the user with interactive involvement of the userend than without.