A new identification methodology, the variable structure system-based identification method, is proposed for multi-input multi-output systems. The developed identifier can successfully estimate unknown system parameters for both linear and some kinds of nonlinear systems that may not be linear in parametric space. The new identifier is essentially a closed-loop system with variable structure control. One major advantage of the proposed identification algorithm is that it provides effective identification for those nonlinear plants if only they are linear in parameter space while in sliding mode, in contrast to other identification algorithms, which usually require that the original nonlinear plants be linear in parameter space. Another advantage of the identification algorithm is that, owing to the robustness of variable structure control, systems with high parameter uncertainties and nonlinearities are not controllable, and consequently identifiable.