To clarify, the drop/recreate of the procedure did likely recompile the procedure. My test was focused on whether or not the changing of the data in the temporary table, and even the drop/recreate of the temporary table would cause the resulting execution plans to differ. It did not.

Your note of 500 or 20% records is important. I worked with a query that I know behaves differently at 2 records versus 50 records in the temp table.

Thanks for the great discussion and forum

ST