This document proposes a new intrusion detection technique that uses data mining algorithms like k-means clustering, neuro-fuzzy models, and support vector machines with a radial basis function kernel. The technique trains different neuro-fuzzy models on subsets generated by k-means clustering, then forms a vector for SVM classification and classifies intrusions using radial basis function SVM. Experimental results on the KDD Cup 1999 dataset show the technique achieves better detection accuracy than other methods like backpropagation neural networks, multiclass SVM, decision trees, and the Columbia IDS model.