During last decades, the public, water authorities and scientists have been more and more concerned by extreme flood events. In the same time, the operational use of flood forecasting and warning systems has increased. However it seems that few systems have been designed in order to take into account short-term time steps. Most available rainfall-runoff models were developed with daily data. Besides, recent work focusing on the coherence of the structures of conceptual models at different time-steps (from daily to pluri-annual) showed the necessity to adapt the model structure to the time-step. This work suggests that models used today at short time steps and developed at larger ones may not be as efficient as they could be, because of inadequate structure and level of complexity. In this context, there is an actual need to carry on this work up to hourly time-step and to develop a lumped conceptual rainfall-runoff model specifically adapted to this time-step. The poster presents the methodology of model development adopted in this research. The design process of the model will be based on evaluation following a methodology developed by Cemagref for a daily model and in comparison with well-known lumped conceptual or empirical models, on a large hydro-meteorological database. The models will be tested on a large sample of catchment, with catchment area typically ranging from 2km² to 2000 km², with low influence of snowmelt and fast response (concentration time inferior to 1,5 day). At least 4 or 5 years of continuous hourly rainfall and runoff series will be needed in each case to get good climatic variability. Ideally, the hydro-meteorological database will include 100 to 200 watersheds, located in as many different countries as possible, in order to test the model in a wide range of hydro-meteorological conditions. The main operational application for an hourly model is flood forecasting, which requires very robust models, with a simple structure. Therefore the model will be based on an existent series of lumped conceptual models. The main objective of this study is to develop the hourly lumped conceptual rainfall-runoff model following : 1. the principle of parsimony, in order to keep the model structure as simple as possible and to avoid the over-parametrization problem; 2. the structure of the model should be coherent with the models developed at larger time-steps.