>John & Franco,
>Rather that just getting a list of those records that have duplicates I actually want the ID of all matching records. So if record 1 matches records 2 and 3 I want to end up with entries so I can compare 1 to 2 and 1 to 3 and 2 to 3

>Id1 Id2
>1   2
>1   3
>2   3

>
>Does that make more sense?

Yes, now I have understood what you need.
I don't thing it's possible with a single select, can you scan/endscan the table?

scan
  "select-sql for the conditions"
  "insert into a cursor the selected id"
endscan

Franco