I'm not aware of any guideline stating "allowed" discrepancies of retention times. And, especially with gradient elution, I would'nt find it any wise. Depending on the "Hardware" (Dwell volume!) and the method itself, rather large differences might be found between different HPLCs. Think of a dwell volume difference of 1 mL - there might be even bigger differences when comparing different HPLCs - and a gradient method with a rather low flow rate, say 0.5 mL. This translates in a dwell time difference of 2 minutes. Assuming that the retention time of your analytes is only dependent on %B - might be oversimplified, but it's justified for this example, i suppose - you will get a retention time difference of 2 minutes, even when using the exact same column and eluents!
I would look at this problem from a practical, pragmatic point of view: Does the retention time difference have any negative effects on the separation? Or to be even more provocing
: Who cares about retention times as long as all peaks are nicely separated
. You're using standards to identify your analytes of interest?
(OK, you might have hard times arguing with an auditor about this...
)
If you are developing methods that are to be run on different HPLCs, you should look at the differences of these machines at an early stage - might prevent you from trouble later.
Transferring methods between different HPLCs CAN be quite troublesome - and sometimes impossible without changes to the methods themselves, which can be really nice
if you're in a regulated environment...