At first glance, I'd guess this is much more CPU intensive than the table comparison technique Tom Kyte pointed out at http://www.oracle.com/technology/oramag/oracle/05-jan/o15asktom.html Both techniques would full-scan the tables, but the HASH_VALUE would have plenty of SQL to PL/SQL context switches. Of course, this is merely an opinion, and I'd certainly be interested in the results (may have to try this one on my own, later tonight.) Glenn On 3/22/06, Jared Still <jkstill@xxxxxxxxx> wrote: > Here's a prototype. > This code is not optimized. :) > <snip> -- //www.freelists.org/webpage/oracle-l