Sean,

Sorry to not have posted that back in the first post, as i stated I though I was missing something obvious enough to discard ddl, i editted it with more information on the problem. On that small set the query optimizer will even identify the rCTE with as the most expensive piece, but when set gets bigger it shifts gears, but again, the rCTE has more IO/CPU on a Set Statistics IO/TIME ON comparison. So my question really is how can it be more IO/CPU time consuming and be rated as more performatic by the Actual Plan percentage?