以下语句完全正确，但COST较高，原因是因为AEH、AEC、AC、ACSN几张表都有数百万条记录， 由于采用了HASH连接，尝试采用优化索引等多种方式，但是由于记录过多，表的查询条件相对较少， COST下降幅度始终很有限，COST提高到1000出头已经很尽力了。 Select Count(Tmp.Id) From (Select Rownum As Id, t.Entry_Id, t.Container_Id, Con.Container_Num              From AEC t              Inner Join AEH Head On t.Entry_Id = Head.Entry_Id              Inner Join AC Con On t.Container_Id = Con.Id And Head.Deleted_Flag = '0' And Con.Deleted_Flag = '0') Tmp        Inner Join ACSN Constatus On Tmp.Container_Id = Constatus.Container_Id And        Constatus.Px_Operation_Status <> '0'         And Tmp.Entry_Id = :B2         And Tmp.Container_Num = :B1  SELECT STATEMENT, GOAL = ALL_ROWS                                               Cost=3519   Cardinality=1       Bytes=62  SORT AGGREGATE                                                 Cardinality=1       Bytes=62   MERGE JOIN                                                                    Cost=3519   Cardinality=51      Bytes=3162    TABLE ACCESS BY INDEX ROWID      Object owner=TEST   Object name=ACSN        Cost=774    Cardinality=49      Bytes=343     INDEX FULL SCAN                 Object owner=TEST   Object name=FK_A_C_C_S_P    Cost=6      Cardinality=1849        SORT JOIN                                                                    Cost=2745   Cardinality=1470    Bytes=80850     VIEW                            Object owner=TEST                           Cost=2744   Cardinality=1470    Bytes=80850      COUNT                         HASH JOIN                                                                 Cost=2744   Cardinality=1470    Bytes=94080        NESTED LOOPS                                                             Cost=1494   Cardinality=1470    Bytes=64680         INDEX FAST FULL SCAN        Object owner=TEST   Object name=IDX_A_E_C   Cost=5      Cardinality=1470    Bytes=33810         TABLE ACCESS BY INDEX ROWID Object owner=TEST   Object name=AEH         Cost=2      Cardinality=1       Bytes=21          INDEX UNIQUE SCAN          Object owner=TEST   Object name=PK_AEH      Cost=1      Cardinality=1           TABLE ACCESS FULL            Object owner=TEST   Object name=A_CO        Cost=1220   Cardinality=1005299 Bytes=20105980   但再细看可以发现有如下苗头：        And Tmp.Entry_Id = :B2         And Tmp.Container_Num = :B1 两个条件是等AEC、AEH、AC三个表HASH连接后才做的过滤，为啥不在连接前就对表做过滤， 再一查在AEH表中有Entry_Id字段，而Container_Num是AC中的一个字段，于是把这两个 条件从对TMP的外层过滤移入到AEH、AC内层过滤中，语句如下；  Select Count(Tmp.Id) From (Select Rownum As Id, t.Entry_Id, t.Container_Id, Con.Container_Num              From AEC t              Inner Join AEH Head On t.Entry_Id = Head.Entry_Id And Head.Entry_Id = :B2 --迁移条件Head.Entry_Id = :B2到此处              Inner Join AC Con On t.Container_Id = Con.Id And Head.Deleted_Flag = '0' And Con.Deleted_Flag = '0'               And con.Container_Num = :B1) Tmp   --迁移条件con.Container_Num = :B1到此处        Inner Join ACSN Constatus On Tmp.Container_Id = Constatus.Container_Id And        Constatus.Px_Operation_Status <> '0'         --And Tmp.Entry_Id = :B2         --And Tmp.Container_Num = :B1  实测结果表明，COST已经下降到了9，提升近500倍，相信这个效果是大家乐意接受的。  SELECT STATEMENT, GOAL = ALL_ROWS                                 Cost=9  Cardinality=1   Bytes=33  SORT AGGREGATE                                                                             Cardinality=1   Bytes=33   NESTED LOOPS                                                                      Cost=9  Cardinality=1   Bytes=33    VIEW                             Object owner=TEST                                                           Cost=7  Cardinality=1   Bytes=26     COUNT                         NESTED LOOPS                                                                   Cost=7  Cardinality=1   Bytes=64       NESTED LOOPS                                                                  Cost=5  Cardinality=1   Bytes=44        TABLE ACCESS BY INDEX ROWID  Object owner=TEST   Object name=AEH             Cost=3  Cardinality=1   Bytes=21         INDEX UNIQUE SCAN           Object owner=TEST   Object name=PK_AEH          Cost=2  Cardinality=1           INDEX RANGE SCAN             Object owner=TEST   Object name=IDX_A_E_C       Cost=2  Cardinality=1   Bytes=23       TABLE ACCESS BY INDEX ROWID   Object owner=TEST   Object name=AC              Cost=2  Cardinality=1   Bytes=20        INDEX UNIQUE SCAN            Object owner=TEST   Object name=IDX_C           Cost=1  Cardinality=1       TABLE ACCESS BY INDEX ROWID      Object owner=TEST   Object name=ACSN            Cost=2  Cardinality=1   Bytes=7     INDEX UNIQUE SCAN               Object owner=TEST   Object name=FK_A_C_C_S_P    Cost=1  Cardinality=1