Summary: The problem of decentralised robust tracking and model following is investigated for a class of uncertain large-scale interconnected systems with both any time-varying delays and any dead-zone input constraints. By combining an adaptive approach with a multiply integral inequality proposed recently in the control literature, a new design method is developed so that (i) the designed decentralised robust tracking control schemes are direct, i.e. not the complicated ones composed of two parts, which can overwhelm the dead-zone input constraints; (ii) it is unnecessary to know any information on the characteristic parameters of dead-zone functions and the time-varying delays, which makes the decentralised robust tracking control rather simple. It is also proved that the local tracking error between the output of each actual subsystem with delays and input constraints and the dynamical signals of the corresponding local reference model is uniformly exponentially bounded. Finally, a numerical example is also given to describe the design procedure and the simulations of this numerical example are implemented to demonstrate the validity of the theoretical results.