The short-term scheduling of a large, complex pumped storage plant is addressed. The corresponding problem is formulated as a mixedinteger nonlinear program where the prices of energy and spinning capacity are assumed known and where two kinds of constraints are considered: dynamic plant constraints, concerning the operation of several units and including startup costs and minimum up time requirements, and also dynamic reservoir constraints, concerning the operation of the upper and lower reservoirs within specified levels. The consideration of dynamic plant constraints together with dynamic reservoir constraints has never been attempted before. It is an extended problem and its solution is difficult to approach. Approaches based on dynamic programming, Lagrangian duality and network flow programming are analysed in view of the extended problem and their advantages and disadvantages are indicated. To bridge the gap between the application range of the programming techniques to this problem of pumped storage scheduling, approximative assumptions and a new approach are proposed. This approach handles the plant constraints by means of dynamic programming, extracts the water values corresponding to the reservoir constraints from the dual solution of a network flow program, and adjusts the water values based on Lagrangian duality. Key results from our numerical experience are presented.