A Hydrothermal Scheduling Algorithm (HSA) for deterministic inflows is presented. A nonlinear network flow model representing the hydrothermal scheduling problem is used. By exploiting the special network structure, a temporally expanded arborescence, the HSA code attains computational savings in processing time and memory requirements allowing the use of microcomputers even for large scale problems. The main feature of the HSA code is a dynamic base choice that enables the decision maker to improve the convergence of the optimization procedure by the use of practical experience. The performance of the algorithm is tested on the Southeast Brazilian Power System.