To forecast riverine floods, short-range forecasts are normally provided. In such cases the initial hydrological conditions highly influence the predictability of a flood event. The study evaluates the potential of an ensemble Kalman filter (EnKF) for the operational flood forecasting system in the Upper Rhone River basin. Observed discharge date is used to update the initial conditions of the hydrological model. Past flood events in the Reckingen subbasin are modelled to assess the robustness of the methodology and the quality of flood predictions.