This paper focuses on a class of large-scale interconnected stochastic nonlinear systems. The interconnections are bounded by strong nonlinear functions that contain first-order and higher order polynomials as special cases. The problem we address is to design a decentralized controller such that the closed-loop, large-scale, interconnected stochastic nonlinear system is globally asymptotically stable in probability for all admissible interconnections. It is shown that the decentralized global stabilization via both state feedback and output feedback can be solved by a Lyapunov-based recursive design method.