Performance　evaluation　is　one　of　the　most　important　standards　to　measure　the　competitiveness　and　productivity　of　enterprises.　Although　existing　studies　could　obtain　the　specific　values　of　enterprises　performance　based　on　historical　data,　they　usually　failed　to　effectively　evaluate　enterprises　performance　in　the　consideration　of　different　indicators.　Meanwhile,　as　the　characteristics　of　existing　performance　evaluation　models　are　uneven,　how　to　choose　a　reasonable　data　envelopment　analysis　(DEA)　model　for　enterprises　performance　evaluation　must　be　considered.　Therefore,　a　new　ensemble　model　on　the　basis　of　homogeneous,　heterogeneous,　and　hybrid　efficiency　evaluation　together　with　the　evidential　reasoning　(ER)　approach　is　proposed　in　this　study　for　enterprises　performance　evaluation,　so　called　the　ER-based　ensemble　model.　The　ER-based　ensemble　model　can　overcome　the　inconsistency　results　caused　by　the　application　of　different　indicators　and　different　DEA　models.　In　case　study,　40　state-own　holding　enterprises　in　China　are　selected　and　all　these　enterprises　are　evaluated　and　ranked　using　the　integrated　efficiency　obtained　from　the　ER-based　ensemble　model.　Comparative　analysis　demonstrates　that　the　ER-based　model　is　better　than　some　traditional　efficiency　evaluation　models　in　enterprises　performance　evaluation　and　performance　ranking.