>>If you need to remove the entirely identical records then what is wrong with :
>>
>>select * dist into newtable
>>then rename and reindex?
>
>That's an interesting idea. Few fields in the table should not be checked (RecNum is one for example). I can select into cursor, zap the original table and append from this cursor.
>
>I'll think about it.

You would not be able to delete any records whose PK is referenced from another table if you have defined any persistent relations [RI]