Abstract: Nonlinear filters are currently used in many applications where the performance of linear filters is unacceptable. The quadratic Volterra filters are introduced as simplest models for the analysis and the identification of nonlinear systems. In this paper, a comparative study of the performances of the LMS and RLS algorithms in the identification of quadratic Volterra systems is presented. A simulation based on a test model is used to show the powerful feature of the RLS algorithm. Moreover, the convergence speed of each algorithm is evaluated according to the variation of the mean-square error criterion on linear and quadratic kernels of the Volterra filters. Also, the effect of noise on the detection feature of the LMS and RLS algorithms are considered in this study.