The supercontinent cycle is the quasi-periodic aggregation and dispersal of Earth's continental crust. There are varying opinions as to whether the amount of continental crust is increasing, decreasing, or staying about the same, but it is agreed that the Earth's crust is constantly being reconfigured. One complete supercontinent cycle is said to take 300 to 500 million years. Continental collision makes fewer and larger continents while rifting makes more and smaller continents.

The most recent supercontinent, Pangaea, formed about 300 million years ago (0.3 Ga), during the Paleozoic era. There are two different views on the history of earlier supercontinents.

The first theory proposes a series of supercontinents: starting with Vaalbara (3.6 to 2.8 Ga); Ur (c. 3 Ga); Kenorland (2.7 to 2.1 Ga); Columbia (1.8 to 1.5 Ga); Rodinia (1.25 Ga to 750 Ma); and Pannotia (c. 600 Ma), whose dispersal produced the continents that ultimately collided to form Pangaea.

The kinds of minerals found inside ancient diamonds suggest that the cycle of supercontinental formation and breakup began roughly 3 Ga. Before 3.2 Ga, only diamonds with peridotitic compositions (commonly found in the Earth's mantle) formed, whereas after 3.0 Ga eclogitic diamonds (rocks from the Earth's crust) became prevalent. This change is thought to have come about as subduction and continental collision introduced eclogite into subcontinental diamond-forming fluids.

The hypothesized supercontinent cycle is concurrent with the shorter-term Wilson Cycle named after plate tectonics pioneer John Tuzo Wilson, which describes the periodic opening and closing of oceanic basins from a single plate rift. The oldest seafloor material found today dates to 170 Ma, whereas the oldest continental crust material found today dates to 4 Ga, showing the relative brevity of the regional Wilson cycles compared to the whole-planetary pulses seen in the arrangement of the continents.

The second view, based on both palaeomagnetic and geological evidence, is that supercontinent cycles did not occur before about 0.6 Ga (during the Ediacaran period). Instead, the continental crust comprised a single supercontinent from about 2.7 Ga until it broke up for the first time, somewhere around 0.6 Ga. This reconstruction is based on the observation that if only small peripheral modifications are made to the primary reconstruction, the data show that the palaeomagnetic poles converged to quasi-static positions for long intervals between about 2.7–2.2 Ga; 1.5–1.25 Ga; and 0.75–0.6 Ga. During the intervening periods, the poles appear to have conformed to a unified apparent polar wander path.

The paleomagnetic data are adequately explained by the existence of a single Protopangea–Paleopangea supercontinent with prolonged quasi-integrity. The prolonged duration of this supercontinent could be explained by the operation of lid tectonics (comparable to the tectonics operating on Mars and Venus) during Precambrian times, as opposed to the plate tectonics seen on the contemporary Earth. However, this approach is widely criticized as an incorrect application of paleomagnetic data.

It is known that sea level is generally low when the continents are together and high when they are apart. For example, sea level was low at the time of formation of Pangaea (Permian) and Pannotia (latest Neoproterozoic), and rose rapidly to maxima during Ordovician and Cretaceous times, when the continents were dispersed.