I bought this lens a few days ago for my new D80. Here are some observations after testing the lens both indoors and outdoors.
It is soft at F2.8, even at the centre, especially at the 50mm setting. At F2.8, performance is better at the wider focal lengths.
Generally, performance seems to improve quite a bit by F4, gets noticeably better by F5.6 and improves even more by F8. I have not printed any of the photographs yet - so do not know what the acceptable maximum print size would be at each aperture/focal length.
For majority of general photography, I would probably stick to apertures F4 and slower, and only use F2.8 if the alternative would be to not take a picture at all.
Build quality seems good - although there are two tiny specs right at the border of the front lens - so small that they are insignificant, but I was not expecting a brand new lens to have this. It was shipped to me direct from Sigma UK - I was surprised at how poorly it had been packed for shipment.
Focusing is silent. Focusing speed is okay for my purposes. There is no manual override in AF mode - despite the fact that this lens uses an HSM.
A few samples taken at f/2.8 and f/8 are posted here. These show the relative performance wide open and stopped down. You will notice that at 50mm, there is a marked degradation in performance at f/2.8.
It was agonizing to choose between this lens, the Tamron 17-50mm, and the Nikkor 16-85mm. I ultimately decided on the Sigma on the basis of better build quality than Tamron, and faster speed than 16-85mm Nikkor. Sadly, the fastest aperture is not as usable as I would have liked.
I think that the tests and reviews available are often misleading (especially Popular Photography review of this lens) - as they do not reflect the real performance of this lens at F2.8. The review that I would say is reasonably accurate is: