Should I worry about the performance of a query like:
SELECT CustomerID FROM invoices WHERE InvoiceID NOT IN (SELECT TOP 50% InvoiceID FROM invoices)
What if the "invoices" is very big?

i would not worry about performance until you can define what "very big" means

also, in the example you gave, you really should have an ORDER BY in the subquery, otherwise it doesn't make sense to pull 50% of all the invoices randomly

The query I used for exemplification doesn't make much sense. My real query it's totaly different but I'm affraid that the inner SELECT will return too many records (hundreds of thousands maybe).

Since I can only see six invoices in your imaginary table, I really doubt that performance will be an issue for that syntax.

-PatP

maybe the inner subquery will be optimized to a join, who knows without at least setting it up and running an EXPLAIN

but a few hundred thousand is typically a piffle

been wantin' to use that word for ages