A new adaptive actuator failure compensation scheme using output feedback is presented in this paper for an enlarged class of nonlinear systems with state-dependent nonlinearities, which are bounded by both linear parameterized static functions and dynamic signals. By introducing the dynamic bounding technique to construct a backstepping control law, the control scheme ensures asymptotic output regulation and closed-loop stability via a Lyapunov based design in the presence of unknown actuator failures as well as system parameter and dynamic uncertainties. Simulation results are included to illustrate the control effectiveness.