In the context of a national energy company (EDF:Electricité de France), hydro-meteorological forecasts are necessary to ensure safety and security of installations, meet environmental standards and improve water resources management and decision making. Given that the actual quality of meteorological and hydrological forecasts do not allow decision-making in a certain future, meteorological and hydrological ensemble forecasts allow a better representation of forecasts uncertainties. Ensemble forecasts improve the human expertise of hydrological forecasts, which is essential to synthesize available information, coming from different meteorological and hydrological models and human experience. In this context, the good estimation and communication of hydrological forecasts uncertainties is an essential step to improve the efficient use of forecasts by end-users. This communication is based on operational experiences and focuses on the estimation and communication of uncertain hydro-meteorological forecasts. First, an operational hydrometeorological ensemble chain developed at EDF is introduced. This chain takes into account both meteorological and hydrological uncertainties, in order to achieve a good probabilistic calibration of forecasts. Probabilistic calibration is absolutely necessary to avoid misrepresentation of uncertainties and under-confidence of forecasts by forecasters and endusers. Then, typical case-studies based on rare hydro-meteorological events will illustrate forecasters' difficulty to estimate and communicate forecast uncertainties. Examples on the Durance (Alps) and Loire (Cevennes) watersheds show the cascading and mixing of uncertainties. Forecasters are used to face rather complex situations and cope with uncertain spatio-temporal meteorological forecasts, uncertain rainfall-runoff models and their own expertise. This communication illustrates the daily forecaster experience of hydrometeorological uncertainties and the difficulties to achieve a good estimation and communication of uncertainties. However, these examples reveal the interest of probabilistic forecasts, compared to deterministic forecasts. Our experience also show that improvements still have to be done, in the field of ensemble forecasts chains, human expertise and communication of uncertainties. On the Durance and Loire examples, meteorological ensembles have both suffered from a strong under-estimation of rainfall risks. However, depending on forecasters experience and confidence, forecasters have been able to correctly estimate and communicate forecasts uncertainties on the Durance and failed on the Loire.