In this paper, the robust neuro-fuzzy networks (RNFNs) are proposed to improve the problems of neuro-fuzzy networks (NFNs) for modeling with outliers. Firstly, the support vector regression (SVR) approach is applied to obtain the initial structure of RNFNs. Because of the SVR approach is equivalent to solving a linear constrained quadratic programming problem under the fixed structure of SVR, the RNFNs are easy to determine the parameters of promise parts and fuzzy singleton of consequence parts. Secondly, when the results of SVR are as initial structure of RNFNs, the annealing robust backpropagation (ARBP) learning algorithm used as the learning algorithm of RNFNs and applied to adjust the parameters of promise parts and fuzzy singleton of consequence parts in RNFNs. Simulation results are provided to show the validity and applicability of the proposed RNFNs.