In the multisensor distributed fusion systems, observations produced by sensors typically arrive at local processors out of sequence. The resulting problem at the central processor/fusion center-how to update current estimate using multiple local out-of-sequent-measurement (OOSM) updates-is a nonstandard distributed estimation problem. In this paper, based on three update algorithms with "Out-of-Sequence" measurement (OOSM), we propose three optimal distributed fusion updates with local OOSM updates, which are, under some regularity conditions, equivalent to the centralized updates with all same lag time OOSMs respectively.