The Short Answer
Earth's mantle is roughly 2,900 kilometers thick. It sits between the crust and the core, and that number comes from seismic data, not from drilling or direct measurement. The crust ranges from about 5 to 70 kilometers depending on whether you are standing under ocean or mountain range, which means the mantle makes up about 84 percent of Earth's total volume. Most people who work with planetary data know this number by heart, but the details behind it are where things get interesting. The exact thickness depends on how you define the boundaries. The top boundary, the Mohorovičić discontinuity or Moho, is relatively straightforward. Seismic waves speed up there because the rock composition changes from felsic continental crust or mafic oceanic crust to peridotite-rich mantle material. The bottom boundary, where the mantle meets the outer core, is much sharper. The Lehmann discontinuity shows a dramatic drop in seismic wave velocity because liquid iron alloy replaces solid silicate rock. That boundary sits at about 2,900 kilometers below the surface, give or take a few tens of kilometers depending on the model you are using. I ran into a practical issue last year while calibrating a seismic tomography dataset. The standard PREM model treats the mantle as a clean layer between two sharp boundaries, but real earthquake data shows the transition zones are messy. The upper-lower mantle boundary around 660 kilometers depth is not a single surface. It undulates by up to 100 kilometers in places, and in subduction zones it can be depressed much deeper. If you are building a model and you assume a flat 660-kilometer discontinuity everywhere, your travel-time predictions will be off by several seconds. The workaround was to apply a regional correction based on the SL2013sv velocity model, which accounts for that undulation using actual seismic observations rather than a theoretical average.
The mantle itself is divided into three main sections. The upper mantle runs from the Moho down to about 410 kilometers. Below that is the transition zone, extending to 660 kilometers, where mineral phase changes shift olivine into denser spinel and then into bridgmanite and ferropericlase. The lower mantle goes from 660 kilometers down to the core-mantle boundary at 2,900 kilometers. Within the upper mantle, the asthenosphere is the softer, partially molten layer that allows tectonic plates to move. It sits roughly between 100 and 200 kilometers deep under oceanic areas and can reach deeper under cratons. The lithosphere, which includes the crust and the rigid uppermost mantle, floats on top of it. One thing beginners consistently miss is that the mantle is not a homogeneous block of hot rock. It has large low-shear-velocity provinces, or LLSVPs, sitting right above the core-mantle boundary. These are massive regions spanning thousands of kilometers across where seismic waves travel noticeably slower. Their exact composition is still debated. Some researchers think they are dense accumulated oceanic crust. Others argue they represent chemically distinct ancient mantle reservoirs. Both explanations fit the data, and neither has been definitively proven. This uncertainty matters if you are using mantle models for anything beyond basic reference, like mantle convection simulations or geochemical tracking. Another counter-intuitive point is temperature. The mantle is solid throughout most of its volume, despite being close to the melting point of its constituent rocks. The geothermal gradient decreases with depth past the crust, and the mantle rocks remain solid because the pressure is too high for melting, except in localized zones where decompression or flux melting occurs. The temperature at the core-mantle boundary is estimated at around 3,500 to 4,000 degrees Celsius. That is hot enough to melt most rocks at the surface, but at those pressures, the solidus is pushed far higher. This distinction matters if you are modeling mantle rheology, because assuming the entire mantle is partially molten would give you completely wrong viscosity values.
If you need to look up mantle thickness values yourself, the USGS and NASA both publish references based on the Preliminary Reference Earth Model, which is the standard. That model lists the mantle thickness at approximately 2,900 kilometers. Other models like ak135 give slightly different numbers, usually within a 20-kilometer range. The variation is small but meaningful for precise geophysical work. There are also limitations you should be aware of. Seismic methods have better resolution under tectonically active regions where earthquake density is high. Under stable cratons and ocean basins with few seismic events, the depth estimates become fuzzier. The Moho depth under the Amazon craton, for example, is estimated at around 80 kilometers, but some studies suggest it could be as deep as 100 kilometers, and the discrepancy comes from sparse station coverage. If you are relying on mantle models for resource exploration or seismic hazard assessment, you should always check the regional resolution of the underlying data. Using a global average model in a poorly constrained area is one of the most common errors I see in published work. The mantle's thickness also varies slightly depending on whether you measure from sea level or from the geoid, and whether you account for the dynamic topography caused by mantle convection itself. Those adjustments are usually small, on the order of a few kilometers, but they matter if you are doing work. Most people asking how thick the mantle is do not need that level of precision, but it is worth noting that the number 2,900 kilometers is itself an average of a somewhat variable reality.
Get the Full Details
