Exploiting the knowledge-based technology and multi-objective analysis, this paper presents a selecting model and its prototype implementation for design for testability (DFT) strategies. Cores to the knowledge-based selecting are decision tree-based knowledge representation models. Keys to the decision tree model are human-like decision procedures and time elimination of defining cost related equations. Test runs over a design-and-test compatible environment demonstrate both feasibility and potential effectiveness of the decision tree selecting model to support both the current and future needs of VLSI testing.