Hi, This should satisfy your requirements: select distinct * from hier connect by prior b =3D a start with a >=3D 100 HTH, Alan Davey -----Original Message----- From: oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx]= On Behalf Of susan lam Sent: Friday, October 15, 2004 8:30 AM To: oracle-l@xxxxxxxxxxxxx Subject: question on hierarchical query hi, I'm having trouble constructing a hierarchical query. Can someone help. SQL> select * from hier; A B ---------- ---------- 99 7 100 1 100 2 1 2 2 3 4 8 200 6 7 rows selected. The order of the output is not important. The output can be: 100 1 1 2 2 3 100 2 2 3 -> should not display this duplicate 200 6 OR 100 1 100 2 1 2 2 3 2 3 -> should not display this duplicate 200 6 We will start with a number. In this case, 100. If a=3D100, get the corresponding value of b and match that back to a and get the next corresponding value of b until b!=3Da. After the first iteration, now get the next value of a>=3D100 and process it in the same manner.=0D Duplicate records should only be displayed once, in this case, 2 3 should only be displayed once. =0D Is it possible to achieve the requirements in a single query? thanks. susan =0D -- //www.freelists.org/webpage/oracle-l "This information in this e-mail is intended solely=0D for the addressee and may contain information=0D which is confidential or privileged. Access to this e-mail by anyone else is unauthorized. If you=0D are not the intended recipient, or believe that=0D you have received this communication in error,=0D please do not print, copy, retransmit,=0D disseminate, or otherwise use the information.=0D Also, please notify the sender that you have=0D received this e-mail in error, and delete the=0D copy you received." -- //www.freelists.org/webpage/oracle-l