The contribution of this investigation lies in solving a sampled-data output feedback stabilization problem for a class of upper-triangular stochastic nonlinear systems in (Formula presented.) -normal structure with unavailable states and uncertain time-varying input delay. A reduced-order observer with the available sampled output detection is developed for the focused system to estimate its unavailable states, and under such observer, a sampled-data output feedback stabilizer is constructed by combining the backstepping and adding a power integral techniques. Through selecting the appropriate Lyapunov–Krasovskii functionals, we can demonstrate that the adverse influences caused by the uncertain time-varying delays in the resulting closed-loop system can be restricted effectively by the constructed stabilizer, and further, by following the stochastic differential theories, the global stability of the considered resulting closed-loop system can be analyzed under the suitable design parameters and allowable sampling period. Finally, two examples are given to verify the effectiveness of the proposed control method.