Looper system of hot strip mills is complex due to characteristics of multivariability, strong coupling, nonlinearity and multiconstrain. Therefore, the decoupling control of looper system has received increasing attention in the field of control. In this study, the decoupling control technology based on PID neural network (PIDNN) was used to eliminate the coupling between the looper height and strip tension, and the particle swarm optimization algorithm was adopted to optimize weights of neural networks. The simulation results show that the proposed control method in this study were effective for the decoupling control. The established method could overcome the problem of nonlinear and strong coupling in wide range, and it exhibited strong robustness.