>>I would give the slight edge to the second, based on the IO Statistics.
>
>Thanks

Just as a follow-up, can you post the structure of the 2 tables, and the expressions for all the indexes (when you use the 2nd scenario on the right)

Even though the second one probably has a slight edge, there's a hash join that's one of the more expensive operators. Wondering if that can be avoided while still getting the two index seeks. I'm not sure, but want to give it a quick look.