This paper is concerned with stability criteria for Ito stochastic functional differential equations (SFDEs) via stochastic differential inequalities. In the paper, the concept "freezing operator" is proposed so that one can extract the associated linear ordinary differential equations from the underlying functional differential equations. Based on the obtained linear ordinary differential equations, a method for constructing Lyapunov functions for SFDEs is proposed in principle. Besides, the quasi-negative definiteness of stochastic derivatives is presented via the freezing operator, and then a new type of stability criteria is established directly based on the quasi-negative definiteness of differential inequalities. As an application, the methods proposed in the paper are applied to establish stability criterion for SFDEs with distributed delays in both the draft terms and the diffusive terms. An illustrating example is finally given to show the method of the paper.