For Flexible Job-shop Scheduling Problem (FJSP) with non-deterministic polynomial characteristics, a cooperative hybrid imperialist competitive algorithm was proposed to minimize the maximum makespan. Firstly, based on process characteristics of standard Imperialist Competitive Algorithm (ICA), improvement of adaptive parameters was designed to improve the convergence speed of the algorithm. Secondly, the reform of the empire and the colonies was introduced. Aiming to the stages of process arrangement and machine selection, a multi-mutation reform strategy was proposed to improve the local search efficiency of the algorithm. Finally, the mechanism for exchanges and cooperation among countries in the mainland was created to promote the exchange of information among outstanding countries and improve the global search capability of the algorithm. By testing on many flexible shop scheduling examples, the experimental results show that the proposed algorithm outperforms many swarm intelligence algorithms in terms of quality and stability, and it is more suitable for solving this kind of scheduling problems.