>>If the original index wasn't filtered, you should have been able to use it for optimization rather than creating two new indexes on your table. Oh, well, at least you have it working now.
>>
>
>I would rather not have to add the 2 additional indexes. I have INDEXSEEK elsewhere on the the index key I already had, so now I have 3 indexes doing the duty of one. Original index expression is simple: custID + poNum. No filter. Should I not need individual indexes?

Is each PoNum unique, or do they overlap? In other words, can customer #1 have po #1, and customer #2 have po #1? If not, then you do not need an index on CustID+PoNum. To get the invoices for a customer:

SELECT * FROM PastInvoices WHERE CustID = ?lcCustID

To get a specific invoice:

SELECT * FROM PastInvoices WHERE PoNum = ?lcPoNum

Hope that helps, rather than confuses :).