Aiming at the scheduling problem for jit delivery under the pulling type production, an optimal method for flexible job-shop scheduling problem (FJSP) based on demand time window is proposed. According to the demand time window of every workpiece, an optimal model for FJSP is constructed to maximize the average membership degree based on the completion time for every workpiece.Then, a tabu search algorithm based on multi-stage hybrid mutation is proposed and designed.To improve the convergence of the algorithm, inverse order mutation and gene segment swap mutation are used to find a good original solution, upon which gene swap mutation is used to find the optimal solution. The effectiveness of the proposed method is validated by applying to a multi-type and small-batch production manufacturing workshop.