The　methods　of　data　stream　mining　have　recently　garnered　a　great　deal　of　attention　in　the　field　of　data　mining,　and　the　sliding　window　technique　has　been　widely　used　during　many　researches　on　it.　This　paper　proposes　a　new　type　of　self-adaptive　sliding　window　(SASW)　model,　which　has　self-adjusting　window　parameters,　and　the　technique　details　are　presented　under　the　ensemble　learning　method　of　single　data　stream　environment.　Experimental　result　shows　that　the　definition　of　evaluating　SASW　parameters　is　appropriate　and　the　gratifying　results　can　be　obtained.　This　idea　also　can　be　used　in　many　other　algorithms　of　data　stream　mining.　©　2013　Springer-Verlag.