Job shop scheduling problem is well known NP-hard problem, this problem is difficult to find the good solution but industrials require the solution to competition and survive in the market. This paper focuses on developing algorithm to solve job shop scheduling problem. The new algorithm is called Future Makespan Heuristic. The concept of new algorithm is jobs are evaluated the makespan before selection one job to the schedule. Each iteration of selection, the new makespan is less than or equal to the makespan of the previous iteration. Iteration makes decision tree by using the candidate jobs (available jobs). Each branch of tree, the candidate jobs are evaluated the makespan by using basic heuristic such as dispatching rules. By this technique, the makespan is improved for all time of scheduling. The experiment reveals the solution is highly improved when comparing with original dispatching rules. Moreover, the computation time of new algorithm is short. The mathematic proof shows that the new algorithm can find the high quality solution in polynomial time algorithm if the evaluating procedure is polynomial time algorithm.