我这有个笨办法，可以尝试下：

SQL> create table a (id varchar2(20),name varchar2(30));

Table created.

SQL> insert into a values('12ab','z3');

1 row created.

SQL> insert into a values('45cd','l4');

1 row created.

SQL> commit;

Commit complete.

SQL> create table b as select * from a;

Table created.

SQL> update b set id='012abc' where id='12ab';

1 row updated.

SQL> commit;

Commit complete.

SQL> select * from b;

ID		     NAME
-------------------- ------------------------------
012abc		     z3
45cd		     l4

SQL> insert into b values ('67g','w5');

1 row created.

SQL> commit;

Commit complete.

SQL> select * from b;

ID		     NAME
-------------------- ------------------------------
012abc		     z3
45cd		     l4
67g		     w5

SQL> create table c (name varchar2(40));

Table created.


declare
 cursor cur1 is select id from a;
 v_id varchar2(100);
 a_id varchar2(100);
 v_sql varchar2(1000);
begin
 open cur1;
 loop
 fetch cur1 into v_id;
 exit when cur1%notfound;
 dbms_output.put_line('v_id is:'||v_id);
 select '%'||v_id||'%' into a_id from dual;
 v_sql := 'insert into c select name from b where id like :1';
 dbms_output.put_line(v_sql);
 execute immediate v_sql
 using a_id;
 end loop;
 close cur1;
end;
/

SQL> select * from c;

NAME
----------------------------------------
z3
l4

可以用select name from b inner join a on instr(b.id,a.id)>0

把表结构和数据拿出来举例

select b.name from a,b where a.id=b.id不行吗？

感觉这是你要的。如果我没理解错的话