AbstractThis paper discusses an optimal short‐range scheduling of multireservoir hydrothermal power systems considering runoff prediction. Using Prasad's rainfall‐runoff model and the Kalman filter‐based, second‐order filtering technique, the hour‐by‐hour characteristic is estimated for a given short‐range forecasted rainfall pattern. Unlike the usual hydroscheduling problem in which a single‐reservoir operating condition is specified at the end of the scheduling period, various end point conditions are examined to deal with the predicted runoff characteristic and to avoid possible water spillages from reservoirs.Dynamic Programming Successive Approximations are applied successfully to obtain optimum operating strategies for many end point conditions at a time. A rapid convergence characteristic coupled with a simple computation algorithm makes the proposed method one of the attractive approaches for scheduling of complicated multireservoir hydrothermal systems considering ru