I'm currently reviewing the new Nikon AF-S Nikkor 28mm f/1.8G and one of the first tasks with any new lens is to check focus accuracy and to test for focus-shift. This particular lens and camera, a Nikon D800, showed consistent front-focus using AF and the combination also revealed focus shift. Both are problematic, though focus shift is the more difficult to deal with (requires conscious effort). To my knowledge there are no Canon or Nikon cameras that make any attempt to compensate for the effects of focus shift (unlike some MF cameras I could mention).
I've used the Michael Tapes Design Lens Align Mk II to demonstrate the focus shift - it's a first class focus calibration tool which was originally designed to highlight issues with DSLRs and AF lenses but it's perfectly suited for this particular purpose. The following images are unretouched JPEGs displayed at 1:1 in Adobe Lightroom 4.
This photo, taken at maximum aperture and focused manually using Live View on the targets (left) at a distance of 1m, shows the zone of focus to be over +6 to -2. This is, arguably, front-focused but it's nothing like the obvious front focus when using AF (the same targets are clearly out-of-focus).
There's no real difference at f/2.0, as you might expect.
Neither is there much difference at f/2.8 save for an increase in DOF; the zone of focus extends from +12 to -4.
By f/4.0 things are starting to get interesting, as there's a shift rearwards (back-focus). The zone of focus covers +14 to -11
It is at f/5.6 that focus shift is most noticeable, with the zone of focus covering +16 to -20. In fairness, the DOF is masking the issue.
At f/8.0 the effects of the rearward focus shift are quite apparent, with the focus zone covering +24 to -36
At f/11 the depth of field masks the focus shift on this scale, although the Lens Align moniker on the scale is clearly out.
At the minimum the DOF extends rearwards so that the chart behind is practically in focus while the front of the ruler/scale is barely in.
These results aren't quite as severe as the focus shift on the Voigtlander 35mm f/1.4 CRF lens tested last month but there is clear evidence of focus shift - we can see the plane of focus move rearwards indicating back focus (under-correction of spherical aberration). This could negatively impact on sharpness with high-resolution cameras, such as the Nikon D800 but I wouldn't call it 'severe', as other internet sites have, as it's mainly masked by DOF. I suspect one of those sites were seeing the effects of focus-inaccuracy rather than focus shift.
In all other respects, the lens performs well. I'll have more to say in a rolling review.
The Nikon AF-S Nikkor 28mm f/1.8G can be bought from Amazon in exchange for $699.