Reservoirs are often operated to meet two or more objectives that sometimes may conflict. An example of this conflict is to keep storage low enough to maintain spare capacity for flood protection, but also to keep it high enough to maintain reliability of water / hydropower supply. Optimising the control decisions of reservoirs can provide a valuable contribution in effectively meeting those objectives. Those decisions require the procurement and interpretation of information from computer models and hydrological data. Any improvement in the availability or accuracy of these sources of information will naturally improve the quality of decision making in operating the water systems. This thesis presents a method for quantifying the value of having different levels of information in operating a test case reservoir. The reservoir selected as a case study is Salto Grande Reservoir, located at the border of Uruguay and Argentina which is operated to maximise hydropower generation while minimising flood risk and damage. The value of those levels of information is done by using a multistage stochastic programming (MSP) on the control of reservoir outflows implemented using a decision tree. There are two key components in making a decision tree. An ensemble of future inflow trajectories, and a representation of information uncertainty along the horizon. The levels of information will influence how uncertainty evolve along the optimisation horizon. Uncertainties constrain the decisions that can be made by a controller. If decisions are constrained, it is likely to be less optimal. Better availability and accuracy of information would result in uncertainty being resolved earlier in the optimisation horizon. The earlier uncertainties can be resolved, the earlier branches in the decision tree can bifurcate from each other, resulting in less constrained decisions. After this methodology is applied to the Salto Grande Reservoir case study, it is confirmed from the results that better information does indeed result in better operational performance. Having accurate forecasts has a quantifiable benefit to the operational performance than having less accurate forecasts or no forecasts at all.