<p>In the research, a two-step identification scheme for the Wiener nonlinear system is developed utilizing correlation analysis and probability density techniques. The Wiener nonlinear system consists of a linear dynamic block and a nonlinear block, wherein the nonlinear block is established through neural fuzzy networks (NFNs), and the linear block is modeled by auto-regressive moving average (ARMA) model. The hybrid signals consisting of Gaussian inputs are introduced to identify separately nonlinear block and linear block parameters. To begin with, based on the distortion-free property of Gaussian signals, i.e., the Gaussian property is preserved when passing through a linear system, the covariance functions characteristics of Gaussian processes are analyzed, then the ARMA model parameter is calculated making use of correlation analysis theory. Furthermore, the probability density function technique is used for making errors tend toward a normal distribution, and is combined with the clustering method to compute unknown parameters of NFNs by utilizing measurable random input–output. The simulation results of numerical and nonlinear process verify that the developed methodology could identify the NFNs-based Wiener system, and reveal superior identification accuracy and control effect.</p>