A new approach to nonlinear system identification using evolutionary neural networks and particle swarm optimization (PSO) has been proposed. Model in this approach consists of a static nonlinear function in series with a dynamic linear system, which has been referred to Hammerstein model. Neural Networks is used to approximate nonlinear function and PSO is resorted to find, the optimal values of the coefficients of the linear part and, the parameters of the nonlinear approximator. Chebychev's polynomials and Taylor's power series are also implemented as two conventional nonlinear function approximators. In this paper despite the other works only input- output data are used to parameters identification and we don't use any other intermediate signals. The simulation results show the effectiveness of NN with respect to latter approximators. Also, numerical example shows that PSO is well suited to deal with system identification problems.