Hi Mikael,
The peak list comparison is, I think, designed to be used with experiments that are the same, e.g. if you want to compare the peaks lists of HSQCs (or other experiments) of your protein with and without a ligand, or at different temperatures or pH values etc.
It sounds to me as though you have a problem that your HSQC peaks don't properly match your 15N-Noesy peaks. There are two possible scenarios.
a) the referencing on one of the spectra is slightly out. The easiest way to check this, is to overlay the two spectra. So make a 3D window in which you have H in x, N in y and C in z. Then you can overlay the two spectra and as you go through your carbon planes you will see whether there is a consistent offset between the spectra or not. If there is a consistent offset, then the best thing is to re-reference the spectra. The easiest way to do this, is to select one peak each in the two spectra which should have the same H and N chemical shifts. Then on one of the peaks do 'Right-click / Re-reference to peak' (check that you are 'moving' the spectrum you want to move, i.e. the one where you think the referencing is more likely to be out).
b) some of your peaks have moved slightly over time and so the HSQC and Noesy spectra are out of sync. I've had issues like this before. Remember that your chemical shifts are averages of all the peaks that contribute to it. If you go to Experiments / Spectra and then the Tolerances tab, you can set the "Shift weighting" for a spectrum. I would just set it to 0 or 0.01 for the HSQC and make sure that for each HN you have at least one assigned peak in your Noesy (only has to be assigned in the HN dimensions). That way your chemical shift list will be based on the Noesy data and not the HSQC data. So during your structure calculation software now has reliable HN chemical shifts for the Noesy spectrum where it is trying to make assignments.
This doesn't exactly answer your question about comparing shift from different spectra, but might help solve the underlying issue.
An easy way to compare shifts between two different sets of spectra could be to go to Experiment / Experiments and put different spectra into different shift lists. You can then compare the shifts lists rather than the peak lists under 'Shift Differences'. You can always put the experiments back into the same shift list again later.
Hope this helps,
Vicky