Gama,
Noise reduction should really only be used for longer exposures to remove dark current pattern noise. If you perform NR on shorter exposures - depending on how NR is performed - you will increase the noise as the camera is read twice and therefore readnoise is increased.
For astrophotography NR is ususally ditched in favour of dark frames, which are generally more effective.
Another point to look for with the D200 is a way to defeat the incamera 'median' filter that is applied to all raw images. Unfortunately, this is carried over from the D50/D70 (see Michael Covington's blog entry for Aug 26, 2005 -
http://www.covingtoninnovations.com/...508/index.html and also Christian Buil's site
www.astrosurf.com/buil ) . In the case of the D70 this can be defeated by turning the camera to NR mode, then switching off the camera as its starts the dark frame phase. Perhaps you can find out. Perhaps with the D200's finer pixel pitch it is not as big a problem.
Lastly, you really dislike Canon don't you! Nearlly every comparision I have seen (Phil Askey, Wayne Cosshall, etc) show the D200 as having a marginal resolution edge, but the 20D/30D's forteit being slightly lower noise at higher ISO's.
I for one would like to see Nikon's become more popular for astrophotography (I wouldn't call myself a brand loyalist, btw I own a holden, and before that a Ford!). However, there are several problems in how they treat raw files that they really need to fix ('median filter', 0 offset, etc). Perhaps in the future they can include a 'scientific mode' that allows access to true raw.
Terry