For complex cooperative control problem of multi-agent systems, a distributed model predictive control scheme based on the dynamic cooperative rules is proposed. The collision avoidance constraint is transformed into hybrid rules based on the positions, and the Boolean function term was introduced in the cost function. In order to accommodate the complex time-varying environment, at each sample moment, the dynamic cooperative rules are designed according to the relative positions among the agents and the relative distances between the agents and the destination, so as to determine the weights in the Boolean function. This scheme reinforces consistencies of motion direction and control action, and improves stability and feasibility of distributed predictive control. As the control target can be achieved via a small prediction horizon, this scheme also enhances the real-time ability and practicality of distributed model predictive control. Simulation examples are given to illustrate the effectiveness of the proposed scheme.