Abstract: To achieve the goal of seismic‐resilient cities, structural systems with post-earthquake functional recoverability have attracted an increasing attention, and self-centering structural systems have become an important research focus in earthquake engineering. Existing studies commonly employ flag-shaped hysteretic models to describe the nonlinear restoring behavior of self-centering structures. However, such models exhibit pronounced nonsmoothed characteristics, and conventional analytical approaches usually rely on smooth equivalent approximations, which may introduce cumulative errors under strong nonlinear conditions. This paper proposes a segmented continuous stochastic response analysis method for nonsmoothed systems. Without introducing any smooth equivalent approximations, the approach proposed improves the traditional exponential polynomial closure method by developing a segmented Gaussian-moment evaluation strategy, which enables the accurate computation of high-order statistical moments for nonsmoothed systems. On this basis, a closed-form solution framework for the evolution of the Fokker–Planck–Kolmogorov equation is systematically established. Compared with conventional equivalent approximation methods, the core advantage of the framework proposed lies in the segmented representation of nonsmoothed restoring forces and of the corresponding efficient and of the accurate evaluation of high-order statistical quantities, without introducing additional approximation errors, thereby significantly enhancing the computational accuracy. Numerical examples demonstrate that the method proposed maintains its high stability and accuracy even under increased seismic intensity, enhanced the system nonlinearity, and more complex energy dissipation mechanisms. The segmented Gaussian-moment framework exhibits a good generality and extensibility, and can be applied to arbitrary nonsmoothed restoring force systems that can be described in a piecewise manner, providing a unified and reliable theoretical tool for the stochastic response analysis of complex nonlinear systems.