I've been wondering about this question for some time now. There are the following two contributors: ~~~~~~~~~~~~~~ 1. Heat left over from the planet's initial formation. In the early 19th century Lord Kelvin estimated the temperature based on a homogenous sphere of uniform initial temperature of 3900K, and estimated (erroneously) 10-100 million years as the age of the earth. Accounting for convection to match observed geothermal gradients pushes the estimate up as much as 3 billion years, using the same initial temperature of 3900K. Planetary accretion generates temperatures in excess of 10000K, although some of this is lost to space. https://websites.pmc.ucsc.edu/~pkoch/EART_206/09-0108/Supplemental/England et 07 AmSci 95-342.pdf 2. Heat generated by radioactive decay. Neutrino studies from Earth's interior, or "geoneutrinos" are validated by comparing their velocity spectra with those emanating from known decaying isotopes in nuclear reactors. Estimates place the total heating power of radioactive decay at about 20TW, about half of the heat flow leaving the earth's interior. https://authors.library.caltech.edu/25422/1/Gando2011p15815Nat_Geosci.pdf ~~~~~~~~~~~~~ So now, depending on the source, I am left with the impression that the geophysical community is not in unanimous agreement about which proportions of these two contributions are responsible for Earth's interior heat. Some of them say that planetary formation is the dominant one, while others say that radioactive decay is more responsible after billions of years. My own (uneducated) opinion is that planetary formation is still the dominant source because the heat from radioactive decay is compared to that merely leaving Earth's interior, not its total heat content. The estimate of 3 billion years is already impressive given that the initial temperature Perry used the same initial temperature as Lord Kelvin, which we now know is far too low. Just asking because I've been really confused about this issue for a while.