Research on the population of Thailand groups it by the family each language belongs to: Tai-Kadai, Austroasiatic, Sino-Tibetan, Hmong-Mien and Austronesian. Genetic structure follows these families, though with wide variation within several of them. Differences between the country's four regions are slight and follow a gradient rather than a boundary, with resemblance to Tai-Kadai-speaking populations of southern China declining from north to south. Regional grouping accounts for under 1 percent of the variation. The largest set of skull measurements from the country gives a similar result, with three regional samples failing to separate from one another.
The origin of Tai-Kadai speakers in mainland Southeast Asia is dated differently in the genetic and linguistic studies. A dated tree of the family's languages places their spread into mainland Southeast Asia within the last 2,000 years and gives the Guangxi–Guangdong coast as the most likely starting point. A study of maternal lineages from nineteen Tai-speaking populations instead places their formation in southern Yunnan. Other genetic studies have examined the question using both maternal and paternal lineages and whole genomes. The mid-20th-century view that made the Yunnan kingdom of Nanzhao a centre of Tai expansion has since been rejected. Ancient genomes associate the arrival of East Asian-related ancestry with the spread of farming. The authors interpret this as admixture with resident populations rather than population replacement. The hunter-gatherer populations present before this transition are genetically closest to the Onge of the Andaman Islands.
Genetic variation is not distributed evenly among the populations studied. Austroasiatic-speaking populations differ from one another by about twice as much as Tai-Kadai populations in repeated sequences, and the difference is even greater in Y chromosome variation. At the other end of the spectrum, the Mlabri are among the most genetically uniform populations studied in the region, while the Maniq of the far south carry maternal lineages that diverged early in the human genetic tree. Evidence of South Asian-related ancestry has also been found in central and southern populations and among the Mon, with estimates placing its arrival between about 500 and 1,200 years ago. These differences have also been relevant to forensic research: reference databases based on Thai samples show differences from neighbouring Asian populations both in the lineages represented and in their frequencies. Similarly, variants of immune-system genes associated with severe drug reactions occur at frequencies that differ substantially from those reported in European and North American reference populations.
Background
Studies of Thailand's populations group them by the family their language belongs to: Tai-Kadai, Austroasiatic, Sino-Tibetan, Hmong-Mien and Austronesian. Tai and Thai are not interchangeable. Tai is one of five branches of the first of those families. That family, also called Kra-Dai, has close to 100 million speakers across South China, mainland Southeast Asia and northeastern India. The reverse does not hold either: the Lao Isan of the northeast hold Thai citizenship and are described as ethnically Lao. How those languages and their speakers reached the country has been read from paternal and maternal lineages, from whole genomes, and from the languages themselves, and the readings do not all agree.
Terminology: Tai, Thai, Siam, and Xian
In English, Tai refers to speakers of languages in the Tai family, most of whom, particularly speakers of Southwestern Tai languages, call themselves Tai or a cognate. Thai now refers to citizens of Thailand and, more narrowly, to the Siamese. Xiān was also used in historical Vietnamese and Chinese sources for polities or peoples in the Chao Phraya basin.
Siam is a loanword rather than a native Tai term. A 2024 study connects it with shan in Burmese, sem in Mon, and siem in Khmer, and argues that these terms were applied by outsiders to speakers of Tai languages who called themselves Tai. Burmese and Mon usage could cover both Thai and Lao, while Shan and Lao could denote political status rather than descent. The linguistic distinction is also reflected in Thai: Siam follows Pali and Sanskrit word order, whereas Thai follows native Thai word order. Before the 19th century, Tai and Lao also had multiple meanings. Tai could refer either to an inhabitant of a place or to a free person, in contrast to a slave, and the Sukhothai inscription traditionally dated to 1292 uses the term in both senses. In the same period, Lao could mean a lord or ruler. The terms may therefore have described social status rather than ancestry or descent.
The earliest attestations of Siam and Thai are uncertain. A 1949 study identified syam on the bas-reliefs of Angkor Wat, dated to about 1150, where Khmer inscriptions refer to mercenaries as syam-kuk. A 1988 study instead dates the term to Old Khmer inscriptions from 611, where it probably denotes a place in the lower Chao Phraya basin and possibly its inhabitants. It cautions that the identity of the syam-kuk mercenaries cannot be established, although they have often been identified as Thai. Thai is securely attested in the Sukhothai inscription traditionally dated to 1292, where it occurs more than twenty times as a name for a people. Tai was not primarily a state name.
La Loubère described Siam as a Portuguese term and reported that the Siamese called themselves Tai rather than Siamese. similarly wrote that foreigners called the kingdom Siam, while its inhabitants knew it as Muang Thai. Both accounts distinguished the northern peoples as Lao; in La Loubère's account, they referred to themselves as Tai Yai. The 2024 study concludes that before the 19th century, Siam, Thai, and Lao were not yet fixed names for nations or ethnic groups, but terms whose meanings varied with context and speaker.
By the mid-19th century, French sources increasingly treated Thai and Siam as referring to the same territory and population; a dictionary published in 1836, for example, equated Siam with Muang Thai. This usage, however, did not extend Thai to Tai-speaking peoples outside the kingdom. A 1949 study dates attempts to extend the term Thai to the Lao and other Tai-speaking peoples to the reign of Chulalongkorn. Siam remained the official name until 1939, when it was replaced by Prathet Thai and Thailand. A 1988 study distinguishes the modern ethnic and linguistic sense of Thai from its legal sense as nationality; the two largely overlap for the majority population, but not necessarily for highland peoples or Malay-speaking Muslims in the south.
Origins and routes
Historical accounts
The account current in the middle of the 20th century, set out in a 1949 survey of the historical record, placed the earliest Tai in the mountains of southern China. It held that Tai-speaking groups moved down the Mekong from about the 12th century, with the Yunnan kingdom of Nanzhao as one of their centres. Sukhothai was regarded as the first Tai kingdom in the region, in the middle of the 13th century, followed by Lan Na in 1292 and Lan Xang in 1353.
The identification of Nanzhao as a Tai kingdom has since been rejected. The idea goes back to Terrien de Lacouperie in the 1880s, who identified the Ai-Lao of western Yunnan as Tai because the Nanzhao rulers claimed them as ancestors, and saw modern Thai and Lao as refugees from the kingdom's fall. The names used by Nanzhao and the vocabulary preserved in Chinese transcription are instead those of Tibeto-Burman speakers. The main population of the Dali basin also speaks a language unrelated to Tai, while evidence for Tai languages in the region begins only in the 11th century. A review of the evidence concludes that any Tai present there played no part in founding the kingdom, places the Tai homeland in Guangxi, and sees Nanzhao as an obstacle to movement north rather than a starting point.
Contemporary records give the name a much earlier history than that account allows. The term syam occurs in Old Khmer inscriptions from the pre-Angkorian period, with the earliest securely dated example from 611. The name appears again in foreign records in the 12th century: the Đại Việt court recorded Xian () in 1149, while the Chinese Lingwai Daida of 1178 records San-lo (), an early transcription that has been linked to the same name. What Xian referred to is itself disputed. A 1988 study reads it as probably Sukhothai, while a later study rejects that identification and places it among the port-polities of the Gulf of Thailand, Ayodhya, Suphanburi and probably Ligor. The dynasty's own account of its origins survives in written instructions prepared in 1684 for a Siamese embassy to Portugal. The envoys were instructed to say that the reigning king descended from a king who ruled in 1300 of the Buddhist era, AD 756 or 757.
Later research places the spread of Tai-Kadai speakers into mainland Southeast Asia before the 12th century. A dated tree of the family's languages places this movement within the last 2,000 years. Complete maternal genomes from Thailand and Laos likewise place it within the last 1,000 to 2,000 years. Ancient genomes from the region show that, by about 2,000 years ago, a component of East Asian ancestry had appeared that was absent from late Neolithic remains. The study reporting this finding interprets it as evidence for the introduction of the ancestral Kra-Dai languages to the mainland.
Linguistic dating
A dated family tree of the Tai-Kadai languages has also been built from word lists rather than genetic data. Comparing 646 sets of related words across 100 languages, it places the first split in the family at about 4,000 years ago, with an estimated range of roughly 2,700 to 5,500 years. The most probable starting point for the spread is the Guangxi and Guangdong coast of southern China rather than the inland provinces. The coastal region has a 47 percent probability, compared with 19 percent or less for each of the four other locations tested. The spread into mainland Southeast Asia is dated to within the last 2,000 years. The same study compares these language splits with an increase in population size estimated from maternal lineages in Tai-Kadai-speaking samples, 266 of which came from Thailand. It links the two to a cold period between about 4,400 and 3,500 years ago. The authors also note that the earlier spread of farming into mainland Southeast Asia is associated with Austroasiatic speakers rather than Tai-Kadai speakers.
Proposed origins
Five maternal lineages found across Thailand and Laos have been proposed as having arisen in the area itself: B5a1d, B6a, R22, R9b and F1f. A maternal survey of 719 people in 19 Tai-speaking populations, from southern China to northeastern India, gives a different starting point. Three lineages carried by about a third of that set, F1a, M7b and B5a, have their root types concentrated in southern Yunnan. Its authors read the Tai as having formed there after descending from populations of southern China. The people sampled in Thailand, Myanmar, Cambodia and northeastern India sit mostly on the outer branches of those lineages, which they read as later arrival. None of its Thai samples was collected for it.
Relationship to Austronesian groups