Recognition　of　acceptor　sites　is　a　part　of　splice　sites　recognition　and　is　an　important　step　in　gene　recognition.　In　this　paper,　We　develop　motif　models　for　acceptor　site　and　a　pattern　match　algorithm.　Using　the　Motif　models,　we　can　mine　out　the　pattern　information　from　acceptor　site　sequences.　The　recognition　results　show　that　this　model　can　correctly　recognize　88.50%　of　the　total　true　acceptor　sites　and　91.50%　of　the　total　false　acceptor　sites,　the　correlation　coefficient　is　0.8004.