A tracking control strategy of unknown nonlinear systems is presented here. Two kinds of neural networks (state-space neural networks and recurrent dynamic neural networks) are used for identification and learning the control law. A sufficient condition, which guarantees the error convergence of the closed-loop system, is given by Lashalle invariant set theorem. This condition gives a guideline for choosing the controller parameter. The tracking control strategy is applied to an unknown general nonlinear system, good tracking performance is obtained in the example, and the corresponding convergence condition is checked.