What genetic studies say about the Pashtuns

At a glance

• Qamar and colleagues (2002) tested 93 Pathan men and found no support for a Jewish origin and none for a Greek one. Their own caveat is the assumption that modern men stand for ancient ones. [2]

• Haber and colleagues (2012) found haplogroup R1a1a-M17 in 51.02% of the Pashtuns they tested in Afghanistan, 30.36% of Tajiks, 17.65% of Uzbeks and 6.67% of Hazaras. [1]

• Underhill and colleagues (2015) found that 98.4% of 490 Central and South Asian R1a lineages belong to the Z93 branch, and that its Z2125 sub-branch reaches more than 40% in Afghan Pashtuns. [5]

• Firasat and colleagues (2007) found two Pathan men with a Y chromosome matching a Balkan haplotype, and read it as European origin for "a small proportion" of Pathan Y chromosomes. [3]

• Ancient DNA from the Swat valley (Narasimhan and colleagues, 2019) shows steppe ancestry reaching the north of South Asia in the second millennium BCE. It does not sample Pashtuns by name. [7]

• Y-DNA follows one line of descent, father to son, and says nothing about the rest of a person's ancestors. [1, 2]

What these studies can test

Pashtun tradition gives a genealogy: the tribes descend from Qais Abdur Rashid and his sons, and the Afghans from Afghana, a grandson of King Saul. Origins of the Pashtuns sets that out and summarises Haber's and Firasat's findings in two paragraphs. Below, the studies are taken one at a time, with the numbers each paper prints. The site's own tribe-by-tribe results are at DNA studies.

A genetic study describes how common certain DNA types are in a group of living people and how they relate to types found elsewhere. It cannot confirm a named ancestor such as Qais or Afghana, because a name leaves no mark on DNA. It can test a claim of origin: if a people descend largely from an outside group, that group's characteristic lineages should appear in them at similar frequencies. The earliest studies were built on that test.

Haplogroups have been renamed as the Y-chromosome tree grew. Qamar's 2002 paper numbers them 1 to 28, Haber's 2012 paper uses names such as R1a1a-M17, and Underhill's 2015 paper works with single mutations such as Z93. Each paper's own labels are kept here.

Qamar and colleagues, 2002

The first large survey appeared in the American Journal of Human Genetics. Raheel Qamar, Qasim Ayub and colleagues typed 18 binary markers and 16 short tandem repeats in 718 men from 12 ethnic groups of Pakistan. [2] The Pathan sample was 93 men, collected with the help of the paediatrics department of the Lady Reading Hospital in Peshawar. [2]

The paper counts 42 of the 93 in its haplogroup 3, 15 in haplogroup 2, 12 in haplogroup 28, 10 in haplogroup 1, 6 in haplogroup 9, 6 in haplogroup 26 and 2 in haplogroup 21. [2] Across Pakistan, the authors found frequencies generally similar to those of neighbouring regions and little geographical clustering. [2]

The paper's first table lists what each group is said to descend from. For the Pathans it gives a Jewish origin (citing Ahmad 1952) and a Greek or Rajput origin (citing Bellew and Caroe). [2] The authors treated each claim as a question of admixture: what share of the group's Y chromosomes could come from the named source?

On the Greek claim, modern Greeks have haplogroup 21 at 28%, it was found in 2% of the Pathans, and the Greek admixture estimate for the Pathans was 0%. [2] On the Jewish claim, haplogroups 21 and 9 are common in Jewish populations (about 20% and 36% in the reference they use), both are low in the Pathans, and "no support for a Jewish origin is found", with an admixture estimate of 0%. [2] They qualify it in the same breath: the conclusion is limited by "the assumption that the modern samples are representative of ancient populations". [2]

Sengupta and colleagues, 2006

Sanghamitra Sengupta and colleagues, in the same journal, asked whether Central Asian herders arrived recently enough to explain South Asia's Y-chromosome variation. They typed 69 markers in 728 Indian samples and in a Pakistani set of 176 men that included 20 Pathans. [4] They concluded that "the influence of Central Asia on the pre-existing gene pool was minor", and that haplogroup R1a1 showed "independent recent histories of the Indus Valley and the peninsular Indian region". [4] The paper is about South Asia as a whole, and its 20 Pathans carry little weight on their own. Later work, below, changed the picture it drew of R1a.

Firasat and colleagues, 2007

Sadaf Firasat, Qasim Ayub, Peter Underhill and others went back to the Greek claim with more markers: 89 binary and 16 short tandem repeats in 952 men, of whom 77 were Greeks and 875 were Burusho, Kalash and Pathans, the three Pakistani groups that claim descent from Alexander's soldiers. [3]

They report that the lowest genetic distances to the Greeks were those of the Pathans. Clade E3b1, which was 17% of the Greek sample and rare in Pakistan, turned up in two Pathan men. One shared a 16-marker haplotype with three Greeks, and the other differed from that cluster by a single mutation. [3] A nine-marker version of the haplotype appeared in 53 of 7,897 haplotypes in a worldwide database, concentrated in Macedonia and Greece. [3] Two mutation rates dated the common ancestor of the Pathan and Greek chromosomes to between about 2,000 and 5,000 years ago, a range that includes Alexander's campaign of 327 to 323 BCE. [3]

The conclusion has two halves. The study "seems to exclude a large Greek contribution to any Pakistani population", and it "provides strong evidence in support of the Greek origins for a small proportion of Pathans". [3] The second half rests on two men, and the authors say it is "based on only a few unrelated descendants". [3]

This differs from Qamar's 2002 reading, which found haplogroup 21 in 2% of Pathans and a 0% Greek estimate. Both agree that Greek ancestry is not large. They differ on whether a trace can be shown, and Firasat used 89 binary markers against Qamar's 18. [2, 3]

Haber's 2012 study looked at the same marker, E1b1b1-M35, in Afghanistan. It was in Hazaras and Uzbeks but not in Pashtuns or Tajiks, and the Afghan lineages grouped with Middle Eastern and Iranian ones, not with Balkan ones. [1]

Haber and colleagues, 2012

Marc Haber and colleagues published the first survey that compared the four largest groups of Afghanistan. They genotyped 204 men from Afghanistan with 52 binary markers and 19 short tandem repeats and compared them with 8,706 samples in all. Each participant had at least three generations of paternal ancestry in his country of birth. [1] The results that bear on Pashtuns, as the paper prints them:

• R1a1a-M17 was in 51.02% of Pashtuns, 30.36% of Tajiks, 17.65% of Uzbeks and 6.67% of Hazaras. [1]

• C3-M217, mainly found in East Asia, was in 41.18% of Uzbeks and 33.33% of Hazaras, against 3.57% of Tajiks and 2.04% of Pashtuns. [1]

• L-M20, H-M69 and R2a-M124, which the authors call autochthonous to India, together made up 20.41% of Pashtuns and 19.64% of Tajiks, against 5.88% of Uzbeks and 5% of Hazaras. [1]

• With all Afghan groups pooled, R1a1a-M17, C3-M217, J2-M172 and L-M20 were most frequent and together made up over 66% of the chromosomes. [1]

The Pashtun and Tajik samples sat closer to northern and western Indian populations than to the Hazara and Uzbek samples. [1] The authors dated the split of the Afghan groups from Iranians, Indians and East Europeans to about 10.6 thousand years ago (95% confidence interval 7,100 to 15,825), and the split of the Pashtuns from the other Afghans to about 4.7 thousand years ago (95% confidence interval 2,775 to 7,725), warning that these dates "should be treated with care" because the method does not model gene flow. [1]

Haber also argued from diversity. R1a1a-M17 is most diverse in the Indus Valley and declines towards the Pontic-Caspian steppe, where its sub-lineage M458 is common. M458 was absent in Afghanistan, which the authors took to mean that R1a1a-M17 does not support the view that expansions from the steppe brought Indo-European languages to Central Asia and India. [1] Finer data has since tested that reading.

Underhill and colleagues, 2015

Peter Underhill and colleagues genotyped 16,244 men from 126 populations across Eurasia, found 2,923 in R1a-M420, and sorted them by sub-branch. [5] R1a split into two large groups: a European one, defined by the mutation Z282, and a Central and South Asian one, defined by Z93. More than 96% of 1,693 European R1a samples belonged to Z282, and 98.4% of the 490 Central and South Asian R1a lineages belonged to Z93. [5]

Within Z93 the sub-branch Z2125 reached its highest frequencies in Kyrgyzstan and "in Afghan Pashtuns (>40%)", and was above 10% in other Afghan groups and some Caucasus and Iranian populations. Sub-branch M780 was frequent across India, Pakistan, Afghanistan and the Himalayas. [5] The text does not say which Pashtun sample the figure comes from.

Sequencing dated the split of R1a from R1b to about 25,000 years ago and the common ancestor of the R1a-M417 group to about 5,800 years ago (95% confidence interval 4,800 to 6,800). The authors placed the first diversification of R1a "in the vicinity of present-day Iran". [5] The dominant Pashtun Y-chromosome lineage therefore belongs to the Asian branch, not the European one. The paper gives no date or route for its arrival among Pashtuns, and does not try to.

The Afghan Hindu Kush study, 2013

Julie Di Cristofaro and colleagues collected 516 blood samples from Hazara, Tajik, Uzbek, Turkmen and Pashtun volunteers. They typed mitochondrial DNA (inherited from mothers) in 90 Pashtuns and the Y chromosome in 87, and ran an autosomal analysis, which draws on ancestors of every line, on a subset that included five Pashtuns. [6]

For Pashtuns they report haplogroup G2c-M377 at 14.7%, almost absent from the other groups; J2a1-Page55 at 3%; and a high frequency of R1a1a-M198/M17, shared only with Kyrgyz. Pashtuns and Kyrgyz showed the lowest Y-chromosome diversity of the groups compared. [6] The authors add that the Pashtuns are "not representative of the other Afghan populations". [6]

On autosomes the Afghan samples as a whole showed "a blend of components from Europe, the Caucasus, Middle East, East and South Asia", each component declining towards Afghanistan, which the authors read as support for a convergence of migrations. [6] Five Pashtuns cannot say more than that about Pashtun autosomal ancestry, and no larger genome-wide Pashtun study is among the sources used here.

Ancient DNA from Swat, 2019

Vagheesh Narasimhan and a large team published "The formation of human populations in South and Central Asia" in Science, with genome-wide data for 523 ancient individuals from Iran, Central Asia, the steppe and northern Pakistan. [7] From the Swat and Chitral districts they used 117 individuals dating from about 1400 BCE to 1700 CE, the "Steppe Cline". [7]

Steppe ancestry reached the north of South Asia in the first half of the second millennium BCE and contributes up to 30% of the ancestry of modern South Asian groups. [7] In the Late Bronze Age and Iron Age Swat individuals only 5% of 44 Y chromosomes were of the R1a-Z93 type found in all steppe herders' males, against 20% steppe ancestry on the autosomes, which the authors read as steppe ancestry entering largely through women. In present-day South Asians they find the reverse, an excess on the Y chromosome, so that steppe lineages reached the ancestors of present-day South Asians mostly through men. [7]

The main text reports no Pashtun groups, and the Swat individuals span a period from long before the Yusufzai settled the valley to the early modern era. For the later history see The Swat valley and The Yusufzai.

The Kalash comparison

The Kalash of Chitral, in the valleys of Bumburet, Rumbur and Birir, are set beside the Pathans in these studies because they too claim Greek descent. [2, 3] Qamar found Greek admixture estimates of 20% to 40% for the Kalash and put it down to drift or poor resolution of the haplogroups, since the Greek marker itself was absent in the Kalash sample. [2] Firasat found only 8 Y lineages among the Kalash against 15 among the Pathans. [3]

In 2015 Qasim Ayub and colleagues genotyped 23 Kalash and sequenced one man. They found a low effective population size (2,319 to 2,603) and no detectable gene flow from neighbours in Pakistan since the Kalash split from other South Asians about 11,800 years ago (95% confidence interval 10,600 to 12,600). [8] Narasimhan's team called the Kalash the group in north-west South Asia with the highest Ancestral North Indian proportion. [7]

The Kalash are a small isolate. Qamar's table gives a census size of 3,000 to 6,000 for them and 17,000,000 for the Pathans. [2] Results for one do not carry to the other. The shared finding is that no study found a large Greek component in either. [2, 3, 8]

What the studies say about the Bani Israel tradition

The Makhzan-i Afghani tells of descent from King Saul through Afghana, and Origins of the Pashtuns gives it in full. The DNA studies speak to it only through what they could have found.

Qamar and colleagues tested the Jewish claim directly. They took haplogroups they described as common among Jewish populations, found them at low frequencies in 93 Pathans, and concluded that "no support for a Jewish origin is found". [2] Haber's paper does not discuss an Israelite origin, but reports that J1-M267, the lineage most common among Arabs, appeared in only one Afghan subject. [1] Firasat's paper, built for the Greek claim, reports nothing on an Israelite origin. [3]

Two limits keep Y-DNA from settling the tradition. It follows only the father's line, and the samples are small: 93 Pathans in Qamar's study against 17,000,000 in its census figure. [2] Qamar's authors state a limit of their own: modern samples may not represent ancient populations. What the studies do show is that no large Middle Eastern component has appeared in any Pashtun Y-chromosome sample so far, and that Haber's Pashtun and Tajik samples lie nearer northern and western India. [1]

Limits of the evidence

• Small samples: Qamar had 93 Pathans, Sengupta 20, Di Cristofaro 87 for the Y chromosome and 5 for autosomes. [2, 4, 6]

• Labels: a man is Pashtun or Pathan by his own statement or the laboratory's record. Haber required three generations of paternal ancestry in the country of birth, and Qamar listed the Pathans as living in the North-West Frontier Province and Baluchistan. [1, 2] Neither sorts men by tribe, so neither can say whether a Yusufzai differs from a Ghilzai.

• Dates come from mutation-rate assumptions and carry wide intervals: Haber's 4.7 thousand years and Firasat's 2,000 to 5,000. [1, 3]

• One line of descent: Narasimhan's team found different male and female histories in the same people. [7]

• Comparison groups: Qamar's estimates assume modern Greeks or Jewish populations stand for the ancient ones. [2]

• Community and commercial tests are volunteers' results, grouped by project organisers and not peer reviewed. The site's DNA studies page draws on such projects and shows R1a at 60% of 104 Ghilzai men, in a chart AfghanDNA posted on 30 August 2026. [9] That is the same direction as Haber's 51.02%, in different men.

How Pashtun readers use these results

Abubakar Siddique, a Pashtun journalist at Radio Free Europe, wrote in 2010 that "modern-day Afghan scholars have rejected" the Mughal historian's writings on origins as biased, since the Mughals and Pashtuns were rivals. He added that one result was a twentieth-century Pashtun counterhistory stressing a connection to Aryan tribes, and that "other scholars have argued that Pashtuns are the Aborigines of Afghanistan". [10] For war-weary Pashtuns, he wrote, "such genetic research is of little consequence": they want peace and development, and to keep their cultural values. [10]

A reader looking for confirmation of any of these views will find each only partly supported. Haber found a shared Afghan heritage reaching back to the Neolithic and a Bronze Age divergence, which concerns men and not a people's name. [1] Underhill places Pashtun R1a on the Asian branch, and Narasimhan finds steppe ancestry in the hills of Swat. [5, 7] None of these touch the question of who first used the name Afghan, which the written record covers in Origins of the Pashtuns.

For a family with a test result, a paternal haplogroup is the line from father to son. It can be compared with the figures above and on the DNA studies page, and the sample sizes there show how much a comparison can carry. A family shajra records names and a test records mutations, and the two answer different questions.