You would have to test it 🙂 but the TOP has to re-perform the query to get the next batch, whereas the Temp Table FROM-TO Range should be using clustered index to locate the rows to be deleted and, assuming the temp table is sorted on clustered key [in the table to be deleted, not the Temp Table's ID range!] then all keys being deleted should be "close" within the physical file which I hope makes the delete process slicker. Still got to dash around all over the disk deleting any associated non-clustered index entries though ...