A novel control algorithm for control of nonlinear discrete-time systems with input delay is presented in the paper. The proposed algorithm is based on adaptive dynamic placement of closed loop system's poles. A controlled nonlinear system is identified by linearizable neural network based simplified additive NARX (NN-SANARX) model and linearized by dynamic output feedback with linear reference model represented by a discrete-time transfer function. Reference model defines the desired zeroes and poles of the closed loop control system. In the proposed algorithm poles and therefore parameters of the reference model are defined as a function of control error providing better characteristics of the transient process. The effectiveness of the proposed algorithm is demonstrated on numerical example.