Abstarct: important with the extension of power networks and connection of local networks to each other. Units planning in power system needs precise load forecast so nowadays load prediction is one of the fundamentals of network management. As load forecasting is itself one sample of function prediction, use of intelligent methods like neural networks and fuzzy systems may help in precise load prediction. The aim of this thesis is to prepare a software which can predict Mashhad The optimal power network management becomes more and more load in an intelligent way. To gain that, back propagation neural network, gradient decent fuzzy system and RLS fuzzy system was tested. Accuracy is the criteria in comparison of these methods. The inputs of all these models were historical data and climate information. The results states that back propagation neural network has desirable accuracy in Mashhad load prediction and fuzzy system with RLS learning algorithm can be a good choice but hardware restrictions must be considered.