Based on the foundation of prediction of networks security situation models, this article proposed a method about applying wavelet kernel function network to prediction of networks security situation. Wavelet kernel function network combined with the neural network and the support vector machines merits, which avoid support vector machine (SVM) solving binding second convex programming problem, network scale doesn't happen dimension disasters problem because kernel function is introduced, and its solution is the global optimal solution, so the situation prediction is more accurate. The experiment tests indicated that this method can accurately acquire the situation value prediction results, it has the good situation prediction potency, which provided one new key for prediction of networks security situation.