Buffer management is a complex problem in the operation of production systems. Poor buffer control may cause congestion and may increase work in process and cycle times. This paper proposes a methodology which attempts to reduce the buffer requirements in job shop production systems. The first step of this approach consists of finding a near-optimal schedule. Next a number of right-shift and left-shift movements are performed within the schedule with the objective of reducing the amount of required buffer slots. The initial schedule is calculated with modified simulated annealing algorithm; the second step is essentially changing the start and/or finish operation dates without changing neither the sequence of jobs on each machine nor the value of the objective function. This last step uses machine blockings to further reduce the amount of buffers. The procedure was tested on a number of benchmark problems from the literature. The obtained results show up to 36% in buffer reduction.