Employing the Razumikhin technique, we present several criteria of input-to-bounded state stabilization and bounded-input-bounded-output stabilization in mean square for nonlinear and quasi-linear stochastic control systems with time-varying delay and time-varying uncertainties.