With　the　development　of　network　technology,　link　prediction　based　on　topological　similarity　has　made　remarkable　achievements.　Among　all　findings,　however,　researchers　lay　more　emphasis　on　the　information　of　paths　between　unlinked　nodes,　but　less　on　that　of　endpoints.　After　extensive　research,　we　find　that　the　endpoint　which　possesses　large　and　extensive　maximum　connected　subgraph　is　more　likely　to　attract　other　endpoints.　And　we　can　also　find　that　endpoint　which　has　both　big　degree　and　H-index　possesses　large　and　extensive　maximum　connected　subgraph.　So,　in　this　paper,　we　propose　a　model　based　on　weighted　synthetical　endpoint　influence　of　degree　and　H-index　and　make　extensive　experiments　on　twelve　real　benchmark　datasets.　The　results　show　that　weighted　synthetical　influence　performs　better　in　accurate　link　prediction.　(C)　2019　Published　by　Elsevier　B.V.