We have a winner.
Even though both tables are on the local machine, the number of fields in each table makes a huge difference. When I created tables exactly like the single field cursors and ran the query on these new tables the speed was equivalent to the query on the cursors.
Thank you all for your input.

>Dore,
>
>Is there more than one field in the origional tables? I don't think subqueries are optimizable, so in the first example all of table1 may need to be passed without an index, but in the second example you pulled a smaller file with just one field. The subquery in the second example would still not be optimizable, but there's far less to load and pass.