Hydrothermal systems are operated by economical scheduling to produce minimum cost for thermal generation, subject to hydraulic and thermal constraints, and demand for electricl energy. The operating planning problem is usually solved by decomposing the original problem into long, medium and short-ter scheduling problems, each one considering suitably specific aspcts for its time step and study horizon. In general, uncertainties are modeled in more details in the long-term horizon, while system generation constraints are more detailed in the short-term horizon. In a previous work long-term scheduling was considered using Artificial Neural Network (ANN) . The available monthly and yearly inlet water data for Tigris river at Mosul, also the data of Mosul dam and its power plant was used to verify the suggested method. In this research work we suggest the use of ANN technique to solve the optimal scheduling of hydrothermal prblem for short-term horizon. Four input parameters are used, which are the main important variales affecting the scheduling. They are the available water for the particular day, temperature of the particulare day, types of the week days and season type. The output of the ANN is the 24 houres water distirbution of the particular day. The water data for Tigris river at Mosul, also the data of Mosul dam and its power plant are used to verify the suggested method.