Scheduling is a persistent problem in manufacturing systems and is considered one of the most challenging combinatorial problems. The scheduling problem of flexible job shops where operations are scheduled on the available flexible machines is commonly referred to as FJSP and is considered as an important scheduling problem due to its nearness to advanced manufacturing systems. During production, the machines need setup for processing two different jobs consecutively. Such setup time is called sequence-dependent setup time (SDST). This paper presents a decentralized multi-agent based model for the FJSP-SDST. For optimization, a popular swarm-based algorithm, particle swarm optimization (PSO), is embedded with the multi-agent model to achieve a global solution. An illustrative example has been explained along with the computational results for the ten standard test cases to show the effectiveness and robustness of the approach.