Communication delay and packet loss highly degrade the control performances of networked multi-agent systems (NMAS). This paper addresses the problem by presenting a novel predictive control technique using a time-delay and packet-loss compensation scheme. The principle of the time-delay and packet-loss compensation scheme is employed to remove the effects of network delay and packet loss. Furthermore, another key part of this paper is to analyse the stability and consensusability of close-loop NMAS on the basis of the designed predictive controller. Finally, simulation results illustrate the efficacy of the proposed method.