In this work, the problem of adaptive neural dynamic surface control (DSC) with the minimum adjustable parameters is discussed for a class of multi-input multi-output (MIMO) pure-feedback nonlinear systems with unmodeled dynamics and output constraints. An auxiliary signal designed by the characteristics of unmodeled dynamics is used to handle the dynamical uncertainties. The unknown continuous black-box functions produced in the controller design process are approximated by using radial basis function neural networks (RBFNNs). Based on an one-to-one nonlinear mapping(NM), the MIMO nonaffine nonlinear system with output constraints is transformed into a novel block-structure MIMO nonaffine nonlinear system without output constraints. Based on the transformed system and modified DSC, robust adaptive neural tracking control scheme is developed. Through theoretical analysis, all the signals in the closed-loop system are shown to be semi-globally uniformly ultimately bounded (SGUUB). A numerical example is provided to demonstrate the effectiveness of the proposed design strategy.