Pashtun ancestry sits at one of the best-sampled crossroads in ancient DNA. The Swat valley in northern Pakistan, the historical Gandhara, holds the largest series of ancient genomes in South Asia, and the mountains between the Hindu Kush and the Indus are where three of the deep streams that formed the region meet. That makes a Pashtun genome unusually easy to test formally.
One rule before any numbers. Ancestry is not identity. Being Pashtun is a matter of language, tribe, custom and family, none of which is measured by allele frequencies. What follows describes the ancient populations that Pashtun genomes resemble and in what proportions. It says nothing about who anyone is.
The short answer: Pashtun genomes are modelled from three deep streams: an Iranian farmer-related stream that reached the Indus region by the fourth millennium BC, the Ancient Ancestral South Indian (AASI) stream native to the subcontinent, and a Steppe pastoralist stream that arrived in the second millennium BC. Pashtuns sit near the Steppe-rich end of the South Asian cline, and the Iron Age and historical-period people of the Swat valley are the closest ancient references. No genome-wide study supports an Israelite or Greek origin.
The three streams#
The framework comes from Narasimhan et al. (2019), a Science study of 523 ancient individuals from Central and South Asia.
The first ingredient is an Iranian farmer-related stream. It is named after the Neolithic genomes of the Zagros (see the Iranian post), but the version that matters for South Asia had split from the Zagros farmers by around 10000 BC and spread east on its own. The second is AASI, a deeply divergent lineage native to the subcontinent and related distantly to the Andamanese. No ancient genome of an unmixed AASI person exists (see the South Asian qpAdm guide).
Those two blended in the greater Indus region. Narasimhan and colleagues found the blend in three individuals buried far from the Indus, at the Bactria-Margiana (BMAC) city of Gonur in Turkmenistan and at Shahr-i-Sokhta in eastern Iran, and named it the Indus Periphery cline. Dated to roughly 3300 to 2000 BC, they are the best stand-in for the Indus Valley Civilisation. The BMAC people around them were different, largely Iranian farmer-related with an Anatolian share and little or no AASI, and BMAC ancestry contributed little to later South Asians.
The third stream is the Steppe. Western Steppe Herder ancestry of the Yamnaya type (see the Yamnaya post) crossed into Central Asia with the Sintashta and Andronovo horizon, mixed with local populations, and reached the Swat valley by around 1200 BC. That Central Steppe MLBA form, not the raw Yamnaya profile, is what entered South Asia.
A dated timeline for the Pashtun homeland#
- Before 5000 BC. Iranian farmer-related and AASI-related people are separate populations.
- About 4700 to 3000 BC. They mix in the Indus region. The Indus Periphery individuals carry the result: roughly two thirds to three quarters Iranian farmer-related, the rest AASI.
- About 1200 to 800 BC. The Swat valley grave cultures at Loebanr, Katelai, Aligrama and Udegram: Indus Periphery-related people with additional AASI and a Steppe MLBA share the study estimates near a fifth. This is the earliest direct evidence of Steppe ancestry in South Asia.
- Around 500 BC to 500 AD. Historical-period Swat, including Butkara and Saidu Sharif. The profile stays largely the same under Achaemenid, Mauryan, Indo-Greek and Kushan rule.
- About 400 to 600 AD. Kidarite and Hephthalite rule. Genome-wide evidence for a large new layer in Swat is thin; later samples remain on the local cline. For the steppe side of that story see the Huns and Xiongnu post.
- Present day. Pashtuns fall at the Steppe-rich, AASI-poor end of the modern cline, close to the Kalash, Kho and other northwestern groups.
Where Pashtuns fall on the cline#
Narasimhan et al. describe present-day South Asians as mixtures of two poles: Ancestral North Indian (Indus Periphery plus Steppe MLBA) and Ancestral South Indian (Indus Periphery plus more AASI). Pashtuns, with the Kalash and other Hindu Kush groups, sit as close to the ANI pole as any population sampled.
Round figures from published models put Pashtun ancestry at about half to three fifths Iranian farmer-related, roughly a quarter to a third Steppe-related, and an AASI share between a tenth and a fifth. Those are ranges because proportions shift with the proxies chosen and because Pashtun communities in different valleys differ.
The Swat valley: the closest ancient references#
Loebanr, Katelai, Aligrama, Udegram, Butkara, Saidu Sharif and Barikot together provide well over a hundred ancient individuals spanning roughly 1200 BC to the first millennium AD, from the Pashtun heartland itself. For an Ancient Matches scan, shared segments with individual Swat genomes are a realistic outcome for many Pashtun kits. A match is shared ancestry with a population, not descent from a named person.
What the Y-DNA and mitochondrial DNA add#
Paternal lineages among Pashtuns are dominated by R1a, specifically the Asian branch R1a-Z93 that the Steppe MLBA populations carried. Haber et al. (2012), studying Afghanistan's ethnic groups, found R1a in about half of Pashtun men. The rest is the familiar northwestern repertoire: L, G2a, J2, Q and R1b each from a few percent to somewhat over a tenth, with E and H at low frequency.
Maternal lineages differ. Quintana-Murci et al. (2004) showed that the Pakistani northwest carries a mixture of West Eurasian mitochondrial lineages (H, U, J, T, K, W) and South Asian ones (branches of M and R). A Steppe paternal signal near one half against a much smaller maternal one is the sex-biased pattern seen across South Asia: the Steppe contribution was carried more by men than by women. No haplogroup is a Pashtun marker; R1a-Z93 is as common among some Central Asian and Indian groups.
Israelite and Greek origin stories#
Two traditions are widely told. The Bani Israel tradition holds that Pashtuns descend from the lost tribes of Israel; another links Pashtuns to the soldiers of Alexander the Great.
Neither has support in genome-wide data. Pashtun genomes fit the local cline of Indus Periphery, AASI and Steppe MLBA ancestry, with no Levantine component beyond what the ordinary Iranian farmer-related stream carries into every West Asian population, and no Greek-specific signal. The Y-chromosome lineages cited for the Israelite tradition, branches of J and E, are widespread across West and South Asia and cannot distinguish a Levantine source from an Iranian one. What the data show is more interesting than either legend: a deep local population that absorbed Steppe pastoralists three thousand years ago and then, through empires that came and went, stayed largely itself.
Limitations#
- AASI has no ancient sample. Every model leans on a modern proxy, usually the Andamanese Onge, and small AASI estimates carry more uncertainty than their standard errors imply.
- Afghanistan is nearly unsampled. The Swat series comes from the Pakistani side. Ancient genomes from Kandahar, Nangarhar or the Kabul valley do not yet exist.
- Pashtun sampling is thin. A confederation of many millions is represented by a few hundred published genomes, mostly from Pakistan.
- Language is not in the genome. The arrival of Steppe ancestry is consistent with the spread of Indo-Iranian languages, but a genome does not record language, religion or tribe.
Frequently asked questions about Pashtun DNA#
Are Pashtuns descended from the lost tribes of Israel?#
Genome-wide studies do not support it. Pashtun genomes fit the local mixture of Iranian farmer-related, AASI and Steppe ancestry seen across northwestern South Asia. The Y-chromosome lineages sometimes cited, branches of J and E, are common across all of West and South Asia and do not point to a Levantine source. The tradition is part of Pashtun culture, not a finding of population genetics.
How much Steppe ancestry do Pashtuns have?#
Among the highest in South Asia. Published models put the Steppe MLBA share for Pashtun and neighbouring Hindu Kush groups at roughly a quarter to a third, with the exact figure depending on the proxies used. The paternal side is more Steppe-shifted still: around half of Pashtun men carry R1a, the lineage the Steppe pastoralists brought.
Which ancient genomes are closest to Pashtuns?#
The Iron Age and historical-period people of the Swat valley in northern Pakistan: Loebanr, Katelai, Aligrama, Udegram, Butkara and Saidu Sharif, dated from about 1200 BC to the first millennium AD. They carry the same three streams in similar proportions. Farther back, the Indus Periphery individuals represent the pre-Steppe base.
Do Pashtuns have Greek ancestry from Alexander's army?#
No study has found a Greek-specific signal in Pashtun genomes. Greek settlement in Bactria and Gandhara is historical fact, but the population those settlers joined was already Iranian- and Steppe-related, so any small Balkan contribution would be hard to separate, and none has been detected.
Are Pashtuns genetically Iranian or South Asian?#
Both descriptions are partly right and neither is complete. The largest stream is Iranian farmer-related, but it arrived in the Indus region thousands of years before Iranian-speaking peoples existed, and Pashtuns also carry AASI ancestry that plateau Iranians lack. On a cline they fall between the Iranian plateau and the rest of South Asia, closest to other Hindu Kush groups.
How to model Pashtun ancestry with qpAdm and Global25#
A Pashtun genome is a rewarding qpAdm target because every stream has a real ancient source in Ancestrify's catalog. The distal model uses Iranian Neolithic Farmer (8000 - 5000 BC), Ancestral South Indian (10000 - 2000 BC) and Western Steppe Herder (5000 - 2800 BC), and tests whether Anatolian Neolithic Farmer (8500 - 6000 BC) or Northeast Asian (6000 - 2000 BC) is needed at all. The proximal model uses the local references directly: Loebanr Swat Iron Age (1200 - 800 BC), Gandharan Swat (BC 200 - 300 AD) and Saidu Sharif (25 - 500 AD), with Parthian Iran (BC 250 - 200 AD) or Peninsular South Indian (BC 600 - 600 AD) as the sources that pull a genome off the Swat profile. A model that passes with Swat sources alone, at a p-value above 0.05 with every weight more than three standard errors from zero, is a strong result.
The catalog is specific here rather than broad, and the Swat sources are the reason. The Global25 analysis adds the worldwide reference set, placing a Pashtun genome against Kalash, Tajik, Punjabi and Iranian samples as well as ancient ones, and Ancient Matches reports shared segments with individual Swat genomes. All three are part of the ancestry service.
Sources and further reading#
- Narasimhan, V. M. et al. (2019). The formation of human populations in South and Central Asia. Science, 365, eaat7487.
- Shinde, V. et al. (2019). An ancient Harappan genome lacks ancestry from Steppe pastoralists or Iranian farmers. Cell, 179, 729 to 735.
- Haber, M. et al. (2012). Afghanistan's ethnic groups share a Y-chromosomal heritage structured by historical events. PLoS ONE, 7, e34288.
- Quintana-Murci, L. et al. (2004). Where West meets East: the complex mtDNA landscape of the southwest and Central Asian corridor. American Journal of Human Genetics, 74, 827 to 845.
Editorial note: the hero artwork in this article was generated with AI as conceptual illustration. It does not reproduce a scientific figure, an ancient individual, a real site or a measured migration route.
