In this paper, a coordinated-distributed model predictive control (CDMPC) scheme is proposed for discrete-time, linear, unconstrained dynamic systems. The proposed control scheme incorporates a coordinator that communicates with local CDMPC controllers. With the assistance of the coordinator, the local CDMPC controllers adjust their calculated control actions iteratively to achieve the optimal plant-wide operation. A 'prediction-driven' algorithm is used to coordinate the local CDMPC controllers. Convergence of the prediction-driven algorithm is shown along with a stability analysis of the closed-loop system under coordinated-distributed control. A simulation example is used to illustrate the effectiveness of the proposed coordinated-distributed control scheme.