Wind power prediction of wind farm plays a decisive role in stable electric power system operation. Firstly to introduce fuzzy neural network�s basic principle; secondly to use 60 days numerical weather prediction (NWP) data and power data, from Jun.08 to Aug.08, as the training data of fuzzy neural network to train the fuzzy neural network; finally to use the next 2 days NWP data, from Aug.09 to Aug.10 as input data of fuzzy neural network, to predict the next 2 days output power of wind farm. The training process and prediction result show that fuzzy neural network has fuzzy decision and judgment, and has good self-learning and adaptive ability, which improves the stability of prediction system and prediction accuracy.