To　address　the　lack　of　health　status　identification　and　poor　stability　problems　in　the　rotating　machinery　equipment,　this　paper　proposes　a　new　method　for　health　status　identification　of　rolling　bearing　based　on　SVM　and　improved　evidence　theory.　Firstly,　in　order　to　reflect　the　rolling　health　condition,　we　use　the　empirical　mode　decomposition　(EMD)　to　extract　energy　value　and　the　original　part　of　the　signal　statistics　constitute　characteristic　parameters.　After　that　we　take　them　as　the　input　to　SVM　classifier　for　the　initial　classification.　Then　we　construct　the　basic　probability　assignment　(BPA)　by　the　SVM　classification　results.　Finally,　the　results　of　recognition　are　given　based　on　recursive　dynamic　combining　weight　distribution　and　decision　fusion.　The　experimental　results　show　that　this　method　can　effectively　identify　Rolling　health　status,　which　has　high　recognition　accuracy,　stability,　and　broad　applicability.