The use of probabilistic forecasts is necessary to take into account uncertainties and allow for optimal risk-based decisions in streamflow forecasting at monthly to seasonal lead times. Such probabilistic forecasts have long been used by practitioners in the operation of water reservoirs, in water allocation and management, and more recently in drought preparedness activities. Various studies assert the potential value of hydro-meteorological forecasting efforts, but few investigate how these forecasts are used in the decision-making process. Role-play games can help scientists, managers and decision-makers understand the extremely complex process behind risk-based decision. In this paper, we present an experiment focusingon the use of probabilistic forecasts to make decisions on reservoir outflows. The setup was a risk-based decision-making game, during which participants acted as water managers. Participants determined monthly reservoir releases based on a sequence of probabilistic inflow forecasts, reservoir volume objectives and release constraints.After each decision, consequences were evaluated based on the actual inflow. The analysis of162 game sheets collected after eight applications of the game illustrates the importance of leveraging not only the probabilistic information in theforecasts but also predictions for a range of lead times. Winning strategies tended togradually empty the reservoir in the months before the peak inflow period to accommodate its volume and avoid overtopping. Twenty percent of the participants managed to do so and finished the management period without having exceeded the maximum reservoir capacity or violating downstream release constraints. The role-playing approach successfully created an open atmosphere to discuss the challenges of using probabilistic forecasts in sequential decision-making.