In a comment below an answer to the question What could have made a satellite appear to move side to side while near the horizon? I've said:
Actually a satellite with flat surfaces - especially solar panels - can deviate quite a bit brighter than expected if the angle is just right. But it's an interesting question how a relatively smooth cylinder, covered in diffuse white paint could do that. It could get dimmer, but I'm not sure if a rocket body can get substantially brighter. Even end-on, the end of a tank would likely be rounded rather than flat, but I'll try to find out if a rocket body has the same "flare" vocabulary as a satellite, it's an interesting question.
Approximate visual brightness of orbital "left-overs" is not so much unpredictable as it is just a lot of hard work with little benefit in most cases. However in serious research (object identification, or monitor of changes) it is certainly done in special cases.
Is there any observational data, or reliable, citable links that address possible large brightness excursions of rocket bodies? This should not include half-exploded craft or those with big plumes of vented propellant. Just the white tube-like ones.
I'm thinking large would be 4 or 5 magnitudes.