Modal　parameter　identification　is　the　core　and　key　of　structural　health　monitoring　and　damage　detection,　which　is　crucial　to　evaluate　the　life　and　safety　of　a　structure.　This　paper　proposes　a　novel　method　for　modal　parameter　identification.　At　first,　the　response　signal　is　pre-processed　by　the　band-pass　filter,　and　then　a　series　of　Intrinsic　Mode　Functions　(IMFs)　are　separated　using　the　Empirical　Mode　Decomposition　(EMD)　from　the　measured　response　signals.　Next　the　real　IMF　is　determined　by　the　correlative　coefficient　between　the　separated　IMF　and　the　measured　signal.　Finally,　the　Natural　Excitation　Technique　(NExT)　and　ARMA　model　are　combined　to　identify　structural　modal　parameters　as　soon　as　the　real　IMF　is　obtained.　The　presented　method　is　applied　to　identify　modal　parameters　of　a　7-storey　framed-structure.　The　identification　results　are　compared　with　those　of　other　identification　methods,　namely　HHT　and　NExT/ARMA.　This　research　shows　that　the　approach　proposed　can　extract　modal　parameters　effectively,　and　also　has　excellent　adaptability.