Abstract. This paper investigates the observer-based robust adaptive control problem for a class of stochastic Hamiltonian systems. The systems under consideration relate to parameter uncertainties, unknown state time-delay and input delay. The purpose is to design a delay-dependent observerbased adaptive control law such that for all admissible uncertainties, as well as stochasticity, the closed-loop error system is robustly asymptotically stable in the mean square. Several sufficient conditions are presented to ensure the rationality and validity of the proposed control laws and observers, which are derived based on Lyapunov functional method. Numerical simulations spell out to illustrate the effectiveness of the proposed theories.