Distributed interactive simulation requires high network bandwidth and computation capacity which caused by redundant data when the system’s scale increases. This situation can be alleviated by data filtering to great extent. But the effective filtering scheme requires simulation hosts receiving accurate filtering information and high processing capability, this consumes too much computation resources when systems grow. In this paper, the authors propose a server-based hierarchical filtering scheme. New scheme frees the simulation hosts from the filtering processing and avails system’s scalability. New scheme also exploits the potential locality of distributed interactive simulation, which reduces the amounts of filtering messages exchanged and filtering computation.