An improved whale swarm algorithm is proposed for solving flexible job shop schedule problem(FJSP) with the objective to minimize makespan based on whale swarm algorithm(WSA). First of all, the position representation and distance calculation method of individuals were well-designed based on processing sequence so that the WSA could solve discrete problem such as FJSP directly. Secondly, cooperating search was introduced to develop "better and near" whale swarm with quality and quantity, expanding the moving region of individuals. Finally, variable neighborhood search(VNS) based on critical path was embedded to enhance the local exploitation ability. Numerical experiments and comparisons were conducted against the best performing algorithms reported in the literature. The results validate the effectiveness and efficiency of proposed algorithm.