This paper addresses the problem of composite tracking control and adaptive learning for discrete-time nonlinear uncertain systems in a general normal form. This problem specifies a joint objective of stable tracking control and accurate learning/identifying the associated ideal control strategy simultaneously, in which the “ideal control strategy” is defined to be the tracking controller structure that is typically adopted when the controlled plant’s nonlinear dynamics are precisely known. To this end, a novel adaptive neural network (NN) learning controller is proposed based on the deterministic learning theory. Compared with existing adaptive NN control approaches, the proposed controller is capable of rendering not only stable tracking control, but also accurate learning/identifying the ideal tracking control strategy. Moreover, the learned knowledge can be effectively represented and stored as constant NN models, whose weights are guaranteed to converge to ideal/optimal values. Based on this, an experience-based controller is also constructed to achieve desired tracking control performance without online adaptation, leading to reduced computational cost and improved controlled performance. Numerical simulations have been conducted to demonstrate the effectiveness of the proposed approach.