Abstract: A new network traffic anomaly detection method based on non-parametric statistics of Gaussian kernel function distribution has been proposed in this paper.In addition,the method is different from the current theory and principle of kernel function applications,such as classification,neural network,machine learning and so on.Considering the fluctuation of the network traffic when anomaly occurs,this paper uses the bandwidth of kernel function as the feature value which can significantly reflect the change of network traffic to analyze the network traffic.Compared with other methods,experimental results show that this method can significantly reduce the computational complexity and false detection rate,also improve the detection rate.