The Eskaleut (/ɛˈskæliuːt/ ^ⓘ e-SKAL-ee-oot), Eskimo–Aleut or Inuit–Yupik–Unangan languages are a language family native to the northern portions of the North American continent, and a small part of northeastern Asia. Languages in the family are indigenous to parts of what are now the United States (Alaska); Canada (Inuit Nunangat) including Nunavut, Northwest Territories (principally in the Inuvialuit Settlement Region), northern Quebec (Nunavik), and northern Labrador (Nunatsiavut); Greenland; and the Russian Far East (Chukchi Peninsula). The language family is also known as Eskaleutian, or Eskaleutic.

The Eskaleut language family is divided into two branches: Eskimoan and Aleut. The Aleut branch consists of a single language, Aleut, spoken in the Aleutian Islands and the Pribilof Islands. Aleut is divided into several dialects. The Eskimoan languages are divided into two branches: the Yupik languages, spoken in western and southwestern Alaska and in Chukotka, and the Inuit languages, spoken in northern Alaska, Canada and Greenland. Inuit languages are divided into several varieties. Neighbouring varieties are quite similar, although those at the farthest distances from the centre in the Diomede Islands and East Greenland are quite divergent.

The proper place of one language, Sirenik, within the Eskimoan family has not been settled. While some linguists list it as a branch of Yupik, others list it as a separate branch of the Eskimoan family, alongside the Yupik and Inuit languages.

The Eskaleut languages are among the native languages of the Americas. They are not demonstrably related to the other language families of North America and are believed to represent a separate, and the last, prehistoric migration of people from Asia. The Alaska Native Language Center believes that the ancestral Eskaleut language divided into the Eskimoan and Aleut branches at least 4,000 years ago. The Eskimoan language family split into the Yupik and Inuit branches around 1,000 years ago. More recent classifications find a third branch, Old Sirenik.

Alexander Vovin (2015) notes that northern Tungusic languages, which are spoken in eastern Siberia and northeastern China, have Eskaleut loanwords that are not found in Southern Tungusic, implying that Eskaleut was once much more widely spoken in eastern Siberia. Vovin (2015) estimates that the Eskaleut loanwords in Northern Tungusic had been borrowed no more than 2,000 years ago, which was when Tungusic was spreading northwards from its homeland in the middle reaches of the Amur River. Vovin (2015) concludes that the homeland (Urheimat) of Proto-Eskaleut was in Siberia rather than in Alaska.

Eskaleut does not have any genetic relationship to any of the world's other language families, this being generally accepted by linguists at the present time. There is general agreement that it is not closely related to the other language families of North America. The more credible proposals on the external relations of Eskaleut all concern one or more of the language families of northern Eurasia, such as Chukotko-Kamchatkan just across the Bering Strait. One of the first such proposals, the Eskimo–Uralic hypothesis, was suggested by the pioneering Danish linguist Rasmus Rask in 1818, upon noticing similarities between Greenlandic and Finnish. Perhaps the most fully developed proposal to date is Michael Fortescue's Uralo–Siberian hypothesis, published in 1998 which links Eskaleut languages to Yukaghir and the Uralic languages.

More recently Joseph Greenberg (2000–2002) suggested grouping Eskaleut with all of the language families of northern Eurasia (Indo-European, Uralic, Altaic, Korean, Japanese, Ainu, Nivkh/Gilayak, and Chukchi–Kamchatkan), with the exception of Yeniseian, in a proposed language family called Eurasiatic. Such proposals are not generally accepted. Criticisms have been made stating that Greenberg's hypothesis is ahistorical, meaning that it lacks and sacrifices known historical elements of language in favour of external similarities. Although the Eurasiatic hypothesis is generally disregarded by linguists, one critique by Stefan Georg and Alexander Vovin stated that they were not willing to disregard the theory immediately although ultimately agreed that Greenberg's conclusion was dubious. Greenberg explicitly states that his developments were based on the previous macro-comparative work done by Vladislav Illich-Svitych and Bomhard and Kerns. By providing evidence of lexical comparison, Greenberg hoped that it would strengthen his hypothesis.

Despite all these efforts, the Eurasiatic language theory was overruled on the basis that mass comparison is not accurate enough an approach. In comparative linguistics, the comparative method bases its validity on highly regular changes, not occasional semantic and phonological similarities, which is what the Eurasiatic hypothesis provides.