This paper focuses mainly on decentralized intelligent tracking control for a class of high-order stochastic nonlinear systems with unknown strong interconnected nonlinearity in the drift and diffusion terms. For the control of uncertain high-order nonlinear systems, the approximation capability of RBF neural networks is utilized to deal with the difficulties caused by completely unknown system dynamics and stochastic disturbances, and only one adaptive parameter is constructed to overcome the overparameterization problem. Then, to address the problem from high-order strong interconnected nonlinearities in the drift and diffusion terms with full states of the overall system, by using the monotonically increasing property of the bounding functions, the variable separation technique is achieved. Lastly, based on the Lyapunov stability theory, a decentralized adaptive neural control method is proposed to reduce the number of online adaptive learning parameters. It is shown that, for bounded initial conditions, the designed controller can ensure the semiglobally uniformly ultimate boundedness of the solution of the closed-loop system and make the tracking errors eventually converge to a small neighborhood around the origin. Two simulation examples including a practical example are used to further illustrate the effectiveness of the design method. KEYWORDS decentralized intelligent control, neural network approximation, stochastic nonlinear systems, uncertain high-order nonlinear systems 2780Recently, many further studies have been developed for the control of high-order nonlinear systems. Sun et al 2,3 used the power integrator method to deal with the control for high-order nonlinear systems. The global adaptive control design was investigated in the work of Zhang et al 4 for uncertain high-order nonlinear systems based on a homogeneous domination approach. 5 The decentralized output-feedback control was developed in the work of Li et al 6 for a class of large-scale high-order stochastic nonlinear systems, and in the work of Sun et al, 7 the improved generalization of the homogeneous domination stabilization scheme was studied. However, the above-mentioned strategies fail to realize the tracking control for completely unknown high-order nonlinear systems; therefore, it is necessary to use the approximation-based control methods.As a solution to the system uncertainties, neural networks (NNs) and fuzzy logic systems have been broadly applied to the modeling and control design. [8][9][10][11][12] By using the approximation property of NNs, Wang and Wang 13 proposed a dynamic learning from the adaptive control. In the work of Dai et al, 14 neural learning control with predefined performance was studied for a marine surface vessel. The nonstrict-feedback structure was considered in the works of Zhou et al 15 and Li et al, 16 and the problem of observer-based adaptive fuzzy control was investigated in the work of Wang et al 17 for nonstrict-feedback systems. The stochastic disturbances were considered in other works 18-20...