The　impact　of　temporal　and　spatial　motion　correlation　on　reference　frame　selection　is　distinct　for　different　block　size　modes.　In　this　paper,　we　study　the　statistical　properties　of　video　sequences　and　the　spatial　correlation　in　video　frames　first,　and　then　propose　an　fast　adaptive　multiple　reference　frame　selection　algorithm　based　on　motion　correlation.　It　can　take　out　the　reference　frames　that　are　not　necessary　for　motion　compensation,　and　reduce　the　complexity　of　coding　adaptively　according　to　the　features　of　the　video　sequences.　The　experimental　results　show　that　our　algorithm　can　save　more　than　40%　of　coding　time　on　average　with　negligible　loss.