AmultisensordistributedextendedKalmanfilteringalgorithmispresentedfornonlinearsystem,inwhichthedynamicequationofthesystemandtheequationsofsensor’smeasurementsarelinearizedintheglobalestimateandglobalpredictionrespectivelyandthesuboptimalglobalestimatebasedonallavailableinformationcanbereconstructedfromtheestimatescomputedbylocalsensorsbasedsolelyontheirownlocalinformationandtransmittedtothedatafusioncenter.Ananalysisofthepropertiesofthealgorithmpresentedhereshowsthattheglobalestimatehashigherprecisionthanthelocaloneandsmallerlinearizationerrorthantheexistingmethod.Finally,anapplicationofthealgorithmtoradar/IRtrackingofamaneuveringtargetisillustrated.Simulationresultsshowtheeffectivenessofthealgorithm.