This paper discusses the scalability problem of distributed virtual environment (DVE) centered on area of interest (AOI) model and dead reckoning (DR) algorithm. Previous models improving scalability of virtual environment are analyzed and compared, and a new adaptive message-filtering algorithm based on DR is proposed. The proposed model can be utilized for distributed interactive simulation (DIS) and various network applications such as multi-player games and virtual communities. The most distinguished characteristic of the proposed model is the introduction of flexible threshold mechanism (FTM). It is inefficient that in static dead reckoning (SDR) algorithm, moving avatar straightly and frequently rotating one are uniformly considered. The proposed model overcomes the weak point of established DR caused by a fixed threshold and a strong dependency on rotational behavior of moving avatars. This paper defines a new criterion for update message filtering, named as update lifetime (UL), which is a valid duration time of entity state update. The advantage of dynamic approaches proposed in this paper is the ability of keeping the balance between fidelity and performance in filtering. In experimental results, it was shown that intelligent DR lowered the increment rate of update message traffic over the increase of behavioral complexity of avatar without any significant loss of accuracy.