A novel model-predictive-control strategy with a timeout scheme and p-step-ahead state estimation is presented in this paper to overcome the adverse influences of stochastic time delays and packet losses encountered in network-based distributed real-time control. An open-loop unstable magnetic-levitation (maglev) test bed was constructed and employed for its experimental verification. The compensation algorithms developed in this paper deal with the network-induced stochastic time delays and packet losses in both the forward path and the feedback path simultaneously. With the p-sampling-period delay upper bound, the networked control system (NCS) can also accommodate up to p1 successive packet losses. Experimental results demonstrate the feasibility and effectiveness of this networked real-time control strategy.