In view of load variety, nonlinear and strong coupling of the wind turbine system, and complex mechanism and large amount of computation of traditional blade root load models, a neural network model based on extreme learning machine (ELM) was proposed to predict the blade root loads in this paper. Firstly, the factors that affect the blade root loads were analyzed, and the model inputs were determined by the principal component analysis. The data from National Renewable Energy Laboratory (NREL) was then used to set up training and test sets, and the blade root loads of a 5MW wind turbine system were modelled and predicted by the proposed method. Finally, the model performance was compared with that of the model established based on the support vector machine (SVM). The simulation results show that the ELM model has high training speed and high prediction accuracy, which verified its feasibility and effectiveness.