The　operational　modal　parameter　identification　is　now　very　active　research　topic　in　the　system　identification　of　civil　engineering　structures.　As　only　output　is　measured,　the　modal　parameter　identification　therefore　bases　structural　output-only　data.　Stochastic　Subspace　Identification　is　a　state-of-the-art　technique　to　identify　the　structural　modal　parameters　through　field　operational　measurements.　To　determine　the　system　order,　however,　is　still　an　issue　to　be　addressed　in　the　data-driven　Stochastic　Subspace　Identification.　Based　on　the　concept　of　the　singular　entropy,　the　singular　entropy　increment　and　its　derivative　are　proposed　in　the　paper　to　determine　the　system　order　incorporated　with　the　Stochastic　Subspace　Identification.　The　corresponding　theory　and　algorithm　are　described　in　detail.　A　simulated　4-strory　frame　structure　and　a　simply　beam　tested　in　the　laboratory　are　presented　to　illustrate　the　applicability　and　reliability　of　the　proposed　singular　entropy　based　method　to　determine　the　system　order.　It　is　demonstrated　that　the　proposed　method　is　effective　and　applicable.