The Mediterranean Sea exists because a narrow opening keeps replacing the water that its climate removes.
Evaporation from the basin exceeds the water supplied by rain and rivers. Today, Atlantic water makes up the deficit by flowing east through the Strait of Gibraltar. Restrict that gateway for long enough and the Mediterranean begins to shrink and concentrate its salts.
That is broadly what happened between 5.97 and 5.33 million years ago during the Messinian Salinity Crisis. Water levels fell, normal marine circulation broke down and evaporite minerals accumulated on a scale difficult to picture. Beneath parts of the modern sea, the resulting salt sequence is more than two kilometres thick.
Then Atlantic marine conditions returned. A widely cited 2009 study in Nature modelled the final reconnection as a self-accelerating flood through Gibraltar. Its headline result was that roughly 90 percent of the incoming water could have entered during a runaway phase lasting only a few months to two years.
The geological crisis is real. The catastrophic flood has growing physical support. The two-year figure, however, is a model estimate rather than a duration read directly from a layer of rock.
A sea with a permanent water deficit
The Strait of Gibraltar is only about 13 kilometres wide at its narrowest point, yet it connects a basin stretching roughly 3,700 kilometres from west to east with the global ocean.
Modern circulation through the strait has two main layers. Relatively fresh Atlantic water enters near the surface. Saltier Mediterranean water returns westward at depth.
Without the incoming layer, evaporation would lower the Mediterranean. It would not vanish overnight, and rivers would continue supplying water, but the basin’s negative freshwater balance makes it unusually sensitive to the state of its Atlantic gateways.
Late in the Miocene, tectonic changes around southern Iberia and northern Africa restricted the marine corridors that had linked the two water bodies. The result was not one uniform event across the entire region. It was a long interval of changing connections, water chemistry and basin levels.
What 640,000 years of restriction left behind
The Messinian Salinity Crisis lasted about 640,000 years. Around Mediterranean margins, geologists find gypsum and other deposits formed from concentrated water. Seismic surveys and drill cores reveal much larger evaporite sequences hidden beneath the deep sea.
The crisis is usually divided into stages. Primary gypsum accumulated around many margins after the onset at 5.97 million years ago. Later, large quantities of halite were deposited in deeper basins. During the final Lago-Mare interval, parts of the Mediterranean received fresher water and supported brackish or non-marine environments.
These stages were interrupted by climatic cycles and changing inflow. The Mediterranean was not simply a full sea on one day and a white salt desert on the next.
The salt itself records repeated concentration of brine past the point at which minerals precipitate. Continued limited seawater inflow followed by strong evaporation can deposit more salt than a one-time evaporation of the basin’s original water.
That matters because kilometres of salt do not, on their own, prove that every part of the Mediterranean dried completely or at the same time.
How low did the water fall?
In the most extreme reconstruction, Mediterranean water levels dropped one to several kilometres below the Atlantic. Rivers cut deeply into exposed margins as they adjusted to the lower shoreline. The remaining water occupied separated, highly saline basins far below present sea level.
Evidence for major drawdown includes buried river deposits, deep erosion and the scale and position of evaporites. More recent work across Sicily and the eastern basin also points toward kilometre-scale differences that could support enormous flood cascades.
But the depth and geographical reach of the drawdown remain disputed. A 2025 review in the Annual Review of Marine Science argues that the familiar story of a desiccated Mediterranean and a single catastrophic reconnection is too often treated as established fact.
Some deposits are consistent with evaporites forming under deep water. Fossils and geochemical records can be read as evidence that Atlantic exchange persisted during parts, perhaps all, of the crisis. Other researchers interpret the same broad record as a strongly isolated basin that underwent dramatic lowering.
“Drawdown” is therefore safer than “the Mediterranean became completely empty.” A colossal lowering is central to the megaflood model, but the precise starting level of the final refill is one of its most consequential uncertainties.
The model that produced two years
Daniel Garcia-Castellanos and colleagues began with a major erosional channel buried beneath younger sediment in the Gibraltar region. The feature extends eastward from the former divide and cuts deeply into rock.
The researchers combined seismic profiles, borehole information and numerical modelling. They asked whether flowing Atlantic water could excavate the channel while refilling a low Mediterranean basin.
The model produced a powerful feedback. A small initial overflow eroded its bed, lowering and enlarging the gateway. A deeper gateway admitted more water. Greater discharge then accelerated erosion, which allowed still more water through.
The early trickle could have persisted for several thousand years without contributing much of the final volume. Once the feedback ran away, the dominant transfer became extremely rapid.
In the model, about 90 percent of the incoming water crossed during this brief phase, lasting from several months to two years. At the peak, Mediterranean water level could rise by more than 10 metres in a day.
Those numbers are calculated consequences of the model. No sediment layer contains annual marks counting exactly two flood years. The result depends on reconstructed topography, the assumed starting depth, channel geometry and how readily the bedrock erodes.
A flood that had to cross Sicily
Water entering through Gibraltar would first refill the western Mediterranean. The eastern basin was separated by the Sicily Sill, so the rising western sea eventually had to spill across another barrier.
A 2024 study in Communications Earth & Environment presented unusually direct evidence for that second cascade. Researchers mapped more than 300 asymmetric, streamlined ridges across southeastern Sicily, all oriented consistently with a large northeastward flow.
They also found poorly sorted breccia between Messinian and early Zanclean formations, deformation and sediment injected into underlying layers, and an erosional channel buried beneath the nearby continental shelf.
The offshore channel was about 20 kilometres wide and connected the onshore corridor toward Noto Canyon on the Malta Escarpment. Flow simulations showed how erosion of the shelf channel and canyon could focus and accelerate water moving into the Ionian Basin.
This land-to-sea sequence strengthens the case that a catastrophic flood crossed from the western to the eastern Mediterranean. It is not an independent measurement of the total refill time. Violent flow across Sicily can be real even if the basin-wide percentage or duration is later revised.
Why salt layers can be kilometres thick
Evaporite minerals form in a sequence as water becomes more concentrated. Carbonates precipitate relatively early, followed by gypsum or anhydrite, then halite and finally highly soluble potassium and magnesium salts under the most concentrated conditions.
The Mediterranean deposits are not one clean, continuous block. They include repeated layers, sediment, erosion surfaces and masses moved or deformed after deposition. Thickness varies sharply among margins, shelves and deep basins.
Salt also flows slowly under pressure. Buried evaporites can form domes, ridges and detachments, distorting the original geometry and complicating seismic interpretation.
The great thickness demonstrates an extraordinary and sustained imbalance between inflow and evaporation. It does not uniquely determine whether salt formed at the bottom of deep brine, in shallow pans, or through different settings at different stages.
The sea surface recovered before the ocean did
Even a rapid return of water level would not instantly create the Mediterranean we know. Incoming Atlantic water encountered basins with different salinities and dense residual brines.
A 2022 Nature Geoscience study reconstructed strong salinity contrasts after the Zanclean flooding. Its model suggested that a sill separating western and eastern waters helped maintain extreme stratification.
The surface could therefore approach ocean level while deep water remained isolated. In that reconstruction, diapycnal diffusion needed about 26,000 years to break down the salinity contrast and restore deep ventilation.
This gives the end of the crisis at least two clocks. The dramatic transfer of most water may have taken months or years in the fastest model. The chemical and ecological recovery of the deep basin could take tens of thousands of years.
“Refilled” describes volume and level. It does not necessarily mean mixed, oxygenated everywhere or ecologically normal.
Why geologists still disagree
The most important evidence is unusually hard to reach. Much of it lies beneath kilometres of water, younger sediment and salt. Seismic waves can image buried structures, but salt bends and scatters the signal.
Outcrops around the Mediterranean expose only fragments of a system later altered by uplift, faulting and erosion. A layer in Spain, Sicily or Cyprus may record local water depth or chemistry without representing the entire basin at that moment.
Time resolution is another problem. The transition at 5.33 million years ago can look abrupt in a core even if it lasted centuries. Conversely, a sequence disturbed by erosion may hide the very interval needed to resolve the flood.
That is why several histories remain in play: a nearly desiccated basin refilled by runaway erosion at Gibraltar; a less extreme basin that retained some Atlantic exchange; or a multi-stage transition containing catastrophic floods without requiring one simple basin-wide emptying and refill.
What is solid, and what remains modelled
The Mediterranean experienced a profound salinity crisis from 5.97 to 5.33 million years ago. It deposited a vast evaporite sequence, underwent major changes in water level and chemistry, and then returned to open marine conditions at the beginning of the Zanclean. These conclusions rest on evidence distributed around and beneath the basin.
Large, erosive floods are also increasingly well supported. Gibraltar’s buried channel, deposits in the eastern basin and the ridge-to-canyon evidence across Sicily describe flows beyond anything observable in the modern Mediterranean.
The figure that 90 percent of the final refill happened within two years belongs to one influential physical model. It is plausible within that reconstruction, widely cited and consistent with a powerful runaway flood. It is not a directly dated fact independent of assumptions.
That distinction is not a retreat from the event’s scale. The rocks preserve the salt, erosion and sudden environmental change. They show that the Mediterranean became a radically different body of water and then rejoined the Atlantic. The model tells us how astonishingly fast the final act may have been.