Among DSLR users there are many who debate over and over ad nauseum the advantages and disadvantages of in-lens vs. in-body image stabilization. Most of the arguments are filled with speculations, marketing spin (trying to sow FUD), hopes, and fears rather than facts. I hope the following helps to separate the two.
This is a summary of the known advantages and disadvantages of in-body and in-lens stabilization -- in contrast to speculations, marketing spin, hopes, fears, etc..
In-body stabilization advantages:
1. Stabilizes *any* lens (new, old, camera company's lens, 3rd party lens, AF lens, MF lens, lens mounted with an adapter, etc.).
2. Price of DSLRs with in-body stabilization are similar, sometimes less, than comparable bodies without it (Sony A200, Pentax K20D, Sony A700, Sony A900).
3. Stabilization gets upgraded regularly and is put in new bodies (2005 KM 7D/5D has 1st generation, 2006 Sony A100 has 2nd generation, 2007 Sony A700 has 3rd generation; 2006 Pentax K100D has 1st generation, 2006 Pentax K10D has 2nd generation). Of course, you don't get the advantage of the new generation of stabilization unless you buy a new body. No doubt there are some people that don't upgrade their DSLRs for many years and are still using a Nikon D1, Canon D30, etc., but most enthusiasts and professionals that use DSLRs upgrade every few years (or even more often) because of the fast advance in overall technology (AF, megapixels, speed of operation, new features, etc.).
4. Stabilization is available instantly with no delay.
In-body stabilization disadvantages:
1. Optical viewfinder is not stabilized. DSLRs with liveview (Olympus E-520) do have a stabilized view when using liveview.
In-lens stabilization advantages:
1. Stabilization works with DSLRs and FSLRs.
2. Optical viewfinder is stabilized.
In-lens stabilization disadvantages:
1. Only available in a very limited set of lenses. And sometimes you have to choose between a mediocre optical and/or build quality IS lens or a much better optical and/or build quality non-IS lens.
2. Rarely, if ever, updated. The only case I know of where an IS lens was updated is the Canon 75-300mm f4-5.6 IS lens got updated to the 70-300mm f4-5.6 IS lens. The old one came out in 1995 and was finally updated in 2005, but it didn't help any of the owners of the 75-300mm f4-5.6 IS. The 1998 28-135mm f3.5-5.6 IS is still waiting for an upgrade and so are many other lenses (Canon 70-200mm f2.8L, etc.).
3. In many cases it adds a lot to the cost of the lens (Canon 70-200mm f4L vs. Canon 70-200mm f4L IS, Canon 70-200mm f2.8L vs. Canon 70-200mm f2.8L IS, Canon 18-55mm f3.5-5.6 vs. Canon 18-55mm f3.5-5.6 IS, Canon 75-300mm f4-5.6 vs. Canon 70-300mm f4-5.6 IS, etc.).
4. When the IS system is asleep there is approximately a 1/2 second to 1 second delay when you half-press the shutter release while the IS system initializes and starts to stabilize. If you shoot too quickly you will often get a blurry photo because the IS lens elements are still getting set and are moving around.
Speculations, marketing spin, hopes, fears, etc.
1. Issues such as whether in-lens stabilization is more effective or, at least, more effective for long focal length lenses is not proven. Various tests that I have seen in multiple places seem to leave it still as an open question. I think the only fair thing to say about this is both systems seem to work pretty well. For one person one system may be a bit better and for another person another may be a bit better. Also, a particular generation of one system might be a bit better than a particular generation of another system, but with different generations of each the results might be the opposite. Frankly, I see this whole issue as sort of a wash. Neither system is perfect, neither system is going to consistently in all situations give you exactly 3 stops or 2.5 stops or 4 stops or whatever.
2. Another issue concerns whether in-lens is mechanically more or less reliable than in-body. This is just speculation too. I'm sure inside the various companies they have their confidential information about the MTBF of their sysytems. It isn't likely they will want to share that info with the whole world though. Since most people replace their DSLRs from time to time for many reasons (mostly because of the fast evolution of the technology) the average user (this is my guess) probably never experiences a shutter failure or failure of other moving parts in the body before the DSLR is replaced for some other reason. In-lens IS is, no doubt, quite reliable too, but it also is a mechanical system and has an MTBF. I think that most people hold onto their lenses for a much longer period than their bodies though.
3. Another issue that gets brought up pretty often is what optical degradation occurs because of the IS lens elements in the lens. Klaus Schroiff at Photozone.de believes it does degrage the optical performance though and he does lots of lens testing. I didn't mention it in the above lists because it is sort of speculation, I guess.
4. In-body stabilization seems to have little effect on the size and weight of the body since comparable bodies from the competition without stabilization are not smaller or lighter (in fact, they are sometimes bigger and heavier). In-lens stabilization seems to add size and weight to the lens, at least in some cases (Canon 18-55mm f3.5-5.6 vs. Canon 18-55mm f3.5-5.6 IS, Canon 17-55mm f2.8 IS vs. Tamron 17-50mm f2.8, Canon 17-85mm f4-5.6 IS vs. Sigma 18-125mm f3.5-5.6, etc.). I put this in the speculation category because it is not known definitively that it is the IS mechanism alone that is causing the size and weight increase.
I'm sorry that I used so many Canon examples and didn't say much about Nikon. I have owned 2 Canon DSLRs and I have owned 3 KM/Sony DSLRs (currently an A700) so I am more familiar with those two systems.