First-order variations of eustatic charts (200–400 m.y.) are in agreement with our understanding of the geodynamic processes that control sealevel. By extrapolation, second-order (10–100 m.y.) and third-order (1–10 m.y.) variations are also thought to follow the same rules. However,this assumption may be jeopardized by a closer examination of the Permian–Triassic example, for which climatic and tectonic eustasy fails toexplain the variations of the eustatic charts. During this period, eustatic charts peak down to their lowermost Phanerozoic values and displaysecond-order variations at rates of up to 3 m/m.y., which is inconsistent with the expected eustatic signal during the early fragmentation ofthe Pangean supercontinent and the late Paleozoic melting of ice sheets. Here, we review the possible mechanisms that could explain theapparent sea-level variations. Some of them do modify the eustatic sea level (ESL). In particular, dynamic deflections of Earth’s surface abovesubduction zones and their locations with respect to continents appear to have been the primary controls of absolute sea level as the Pangeansupercontinent formed and broke up. Other mechanisms instead only locally or regionally produced vertical ground motions, either upliftingcontinents or tilting the margins where the control points were located. We show that (1) the thermal uplift associated with supercontinentinsulation and (2) the dynamic uplift associated with the emplacement of a superplume both give rates of sea-level change in the range oflong-term changes of ESL. We also show that (3) the dynamic tilt of continental margins not only produces apparent sea-level changes, butit also modifies the absolute sea level, which in turn may end up in the paradoxical situation wherein fingerprints of ESL drop are found inthe geological record during actual ESL rise. We conclude that second-order absolute sea-level changes may remain elusive for some time.