It seems that some of the posters having experience with different sensors confirm the validity of DR measurements by DxO.
Marc McCalmont has posted some info. He has or have had P45+, IQ180, Canon 5DII, Pentax K5 and now Nikon D800E. His experience essentially agrees with DxO-mark.
Mark Dubovoy, the guy who claimed 6 stops of DR advantage of MFD over DSLRs now seems to be content with the DR on his D800E.
I guess that some of the info regarding DR advantage of MF over DSLRs is coming from Canon users, all Canon cameras are pretty bad on DR at base ISO.
Although it's possible to have good high ISO performance and have bad DR at base ISO, which the Canons illustrate, any camera having good DR at base ISO would also excel at high ISO, because increasing ISO is basically just underexposure. On the high end, near saturation all sensors are created almost equal.
Now, I'd say that we need to look at the whole system, subject, lens-shade, lens, camera body, sensor, tripod and photographer. That still leaves out what is made in post. Phase One also develops Capture one, one the best raw converters. It is quite possible that Capture One can make the best of Phase One images.
This of course assumes that DXOs measure of DR corresponds perfectly to most photographer's needs.
Noise and tonality is a funny thing, two images can have very similar numerical measures of noise and one can be a beautiful aesthetically pleasing gaussian grain and the other a horridly ugly clumping of crappy noise. This is especially when it comes to smooth transitions of tone and color accuracy because when you need to dig deep into shadows it is these attributes that either allow you to include shadow detail as part of the overall frame in a way that looks natural or not.
I'm not saying that is the case here. I've not spent enough time yet with a D800 to comment. I'm simply saying that purely numerical measures of photographic quality are always very dangerous to use on their own.
DXO is a great source of information and I think the photo community is lucky to have them. But they are at their most useful when comparing similar camera systems (e.g. one generation of nikon dSLR to the next generation of nikon dSLR), rather than comparing systems that are different in just about every way (e.g. D800 vs. IQ180). Especially once you through software in the mix where some of the magic of an integrated software/hardware solution like Phase+C1 or Hassy+Phocus comes into play.