Abstract: The technique of dead reckoning（DR） gives a way to reduce the network communication costs and network latency in distributed interactive simulation. But each of the present algorithms of DR owns only one threshold, so it could not differentiate the entity in the beyond from the entity in the nearby. The technique of DR with multi threshold discussed in this paper divides the DOR into several child DOR. Each of local entities reckons differing from the child DOR with different threshold. The technique of DR with multi threshold resolves the conflict between the behavior distortion of the entities nearby and the network communication costs, and it works well.