Constructing　gene　network　and　analyzing　its　structure　is　a　basic　but　important　way　to　discover　and　understand　unknown　physiological　mechanisms　of　Arabidopsis　subject　to　external　stimuli　on　both　the　cellular　and　molecular　levels.　Mutual　information　networks　are　identified　based　on　gene　expression　profiles　of　Arabidopsis　in　normal　condition　and　subject　to　stimuli.　The　comparison　and　analysis　of　these　network　structures　show　that　four　network　statistics,　average　degree,　average　clustering　coefficient,　modularity,　the　proportion　of　non-isolated　nodes　can　distinguish　the　two　types　of　networks　corresponding　to　normal　condition　and　those　subject　to　different　stimuli.　A　new　method　is　proposed　to　classify　these　networks　in　multidimensional　parameter　space.　Our　studies　can　help　better　understanding　of　physiological　mechanisms　of　genome　response　to　stimuli.　©　2011　IEEE.