Though, to play devil's advocate for a moment longer, if you're using a perfectly transparent delay as a plugin/rack unit to a pre-recorded track, you're delaying the sound that came out of the amp speaker and into the mic - in fact, especially at higher poweramp volumes and greater saturation, to do so will actually give you a
clearer delay that's more identical to the dry sound than using a delay in your FX loop, as the delayed signal will not be echoing the poweramp distortion, but will be interacting with your dry signal as it (undelayed) hits your amp's poweramp - i.e - you won't be capturing the dry signal exactly as it is, saturating the poweramp as it would on its own, but rather you're capturing the impact of poweramp saturation from the impact of both the dry and delayed singals hitting at the same time.
I'm just being anal and nitpicky here because it's late, I'm at work, I'm supposed to be studying for something, and my brain is on the fritz, but if your idea is to, in your delayed sound, get exactly the same sound as what would be coming through the amp's speaker and microphone, I still think the best bet is to add delay post-tracking.
...unless you're talking about live sound reinforcement, and adding delay at the amp instead of the board/mains of course, in which case never mind.