I am in the process of comparing files I shot, handheld, of a brick wall, outside a camera store, using my 5D and 17-40, 24-105, 50/2.5 (didn't bother with 70-200/4IS as I assumed that had enough resolution for 5D2 sensor) and their 5D2 with same lenses. Unfortunately store insisted on using their own lenses on the 5D2, except for 17-40 which they didn't have.
Alongside my practical test, I wondered what conditions forum users considered allowed 5D2 sensor gains to be realised, taking as given my most used lenses are 17-40, 24-105, 70-200 f4IS (i.e. not up for changing these) I appreciate first two may realise some but not all sensor benefit, but would you expect to get significant benefit handheld at mid-apertures at the shutter speeds you would normally use to avoid shake, taking IS into account, or is it only realisable on a tripod, with MLU?
Based on my very imperfect tests, with Jpegs (can't read 5D2 raw files), 17-40 looked about the same (19mm, f8@320th), and my 24-105/5D looked better than
their 24-105/5D2 (24mm, f8 1/60th) All 200 ISO, viewed at 100% on screen
As I said very imperfect tests, with lots of variables, but to me indicative that in hand-held street photography I won't see the pixel advantage, only in tripod mounted shots with MLU. Anyway, don't feel I'm persuaded of value in upgrade at net cost of about £1200-1300 sterling, although would love the improved LCD screen and AF micro-adjustment, live view etc.