The　frequency　response　functions　(FRFs)　are　frequently　used　to　identify　dynamic　properties　of　the　mechanical　joints.　However,　the　ill-conditioning　problems　of　the　system　are　caused　by　the　contamination　of　measurement　noise　even　if　it　is　processed　by　reduction　technique.　In　order　to　solve　this　problem　and　make　the　identified　results　represent　real　properties　of　the　joints,　an　improved　FRFs　method　was　developed　based　on　condition　number,　in　which　the　measured　data　were　processed　before　parameter　identification.　The　average　values　of　the　condition　numbers　could　be　found　to　ensure　that　condition　number　is　small　enough.　Moreover,　the　frequencies　corresponding　to　those　condition　numbers　which　are　larger　than　average　one　are　picked　out,　and　comparing　those　retained　ones　the　intersection　frequencies　are　found　under　which　the　parameters　are　identified,　thus　the　ill-conditioning　problem　can　be　avoided.　The　results　of　simulation　and　experiment　show　that　the　proposed　method　can　solve　the　ill-conditioning　problems,　thus　significantly　improve　the　accuracy　of　identification.