In this paper, the cooperative distributed model predictive control (DMPC) problem of a class of complex systems consisting of several subsystems is studied. The states of these subsystems are coupled with each other, and thus bring challenges for the model predictive control algorithm. Moreover, though the subsystems can communicate with each other, they only can access to the output information of their neighboring subsystems. In this case, Luenberger observers are used to estimate the unknown states and a distributed prediction strategy is established for the studied system. Then, the optimal control of the closed‐loop system is realized by designing distributed model predictive controller on the basis of the estimated states. The terminal constraints are introduced in the proposed DMPC algorithm to ensure the iterative feasibility and also the stability of the closed‐loop. Finally, the effectiveness of the proposed method is verified by a numerical simulation.