The first major application of computing machines and systems analysis to the design of water resource systems concentrated on the structural components (dams, turbines, diversion works, etc.) and left aside the difficulties inherent in selecting optimal or near-optimal operating policies. Subsequent work, particularly that which dealt with mathematical programming and its several derivatives, offered new insights into the construction of optimal operating schemes. This paper advances one step further in that it combines the selection of components, the specification of optimal operating policies, and the role of streamflow forecasting in a single mathematical entity. No formalism for forecasting is given, but based on a widely accepted model for flow sequences, a technique is developed whereby the analyst can discern whether or not it is useful to include forecasting in the operating policy. The procedure makes extensive use of statistical theory and Monte Carlo analysis.