Multi-agent simulations model individual behaviors of human or facility with their interaction for predicting emerging collective phenomena. The simulation models have been applied with much attention in various sociological and scientific fields. However, as simulation models get more complicated, there need simulation environments that are capable of executing large size of agents. In this paper, we develop a distributed multi-agent simulation system using HLA (High Level Architecture) and RTI (Run-Time Infrastructure) that are applied for conceptual design and implementation of distributed simulation. Experiments are performed to test that the proposed simulation system can resolve the performance problems occurred from distributed computing networks.