Parsi DNA
Genetic origins and closest populations · India · South Asia
Parsis are Zoroastrians who, according to their own tradition, left Iran after the Arab conquest and landed on the coast of Gujarat, probably between the 8th and 10th centuries AD; many later settled in Mumbai. This page uses the Parsi average from India. The profile points west: in our model, it is 35.9% Iranian Neolithic farmer, 17.4% South Asian hunter-gatherer related (Irula proxy), 16.5% Anatolian Neolithic farmer, 14.9% steppe, 12.1% Levantine Neolithic farmer and 3.2% Caucasus hunter-gatherer. The South Asian hunter-gatherer related share is low for India, similar to the Baloch (15.8%), and Levantine farmer ancestry is rare in South Asian averages. Apart from a Pakistani Parsi sample (0.0056), the closest modern averages are the Qizilbash and Persians from Khorasan in eastern Iran, all above 0.03, so Parsis today stand somewhat apart from both Iranians and Indians. The closest ancient groups are Late Bronze Age people from Dinkha Tepe and Hasanlu in northwestern Iran (0.0608 and 0.0623).
- Largest ancient components in our model: Iranian Neolithic farmers 35.9%, South Asian hunter-gatherer related (Irula proxy) 17.4%, Anatolian Neolithic farmers 16.5%
- Closest modern population in our data: Parsi (Pakistan), distance 0.0056
- Closest ancient group in our data: Iran LBA Dinkha Tepe (Iranian Plateau Profile), distance 0.0608
Where does Parsi DNA sit on the genetic map of the world? To answer, we take the averaged Global25 (G25) coordinates of the Parsi individuals sampled in India, listed in the G25 sheet as "Parsi_India" and compare them with deep ancestral reference populations, with other modern groups and with ancient genomes. It is one of 335 populations in our atlas.
Ancient make-up of the Parsi average
The ancient make-up of the Parsi average is led by Iranian Neolithic farmers (35.9%), followed by South Asian hunter-gatherer related (Irula proxy) (17.4%) and Anatolian Neolithic farmers (16.5%). The Iranian Neolithic farmers component reflects the early farmers and herders of the Zagros mountains in Iran. The South Asian hunter-gatherer related (Irula proxy) share (17.4%) reflects the deep hunter-gatherer ancestry of South Asia (Ancient Ancestral South Indians).
Smaller traces of Caucasus hunter-gatherers (under 5%) also appear. At that level they can reflect real minor ancestry, but also simple model noise, so they should not be over-read. The model fits well (fit distance 0.024), so these proportions are a reasonable summary. This population was modelled with a global set of reference populations (ancient genomes, plus modern stand-ins where no suitable ancient genome exists), because West Eurasian sources alone cannot describe it.
Sindhu – South Asian Report
Closest modern populations to Parsi
Among the modern populations in the G25 data, the closest to the Parsi average is Parsi (Pakistan), at a distance of 0.006 (very close). Next come Qizilbash at 0.032 and Iranian (Persian Khorasan) at 0.034. Leaving aside the other regional samples listed under the same name in the G25 sheet, the nearest other populations are Qizilbash (0.032), Iranian (Persian Khorasan) (0.034) and Persian (Khorasan) (0.038). By the tenth closest (Balochi (Iran)) the distance grows to 0.051, a common pattern for a population that shares ancestry with its neighbours while keeping a profile of its own.
| # | Population | Distance | Closeness |
|---|---|---|---|
| 1 | Parsi (Pakistan) | 0.0056 | Very close |
| 2 | Qizilbash | 0.0318 | Close |
| 3 | Iranian (Persian Khorasan) | 0.0336 | Close |
| 4 | Persian (Khorasan) | 0.0378 | Close |
| 5 | Iranian (Bandari) | 0.0403 | Close |
| 6 | Tajik (Herat) | 0.0406 | Close |
| 7 | Persian (Kerman) | 0.0412 | Close |
| 8 | Persian (Sistan) | 0.0432 | Close |
| 9 | Ajam (Emirates) | 0.0506 | Moderate |
| 10 | Balochi (Iran) | 0.0514 | Moderate |
Distances are Euclidean distances between averaged G25 coordinates. On our scale, below 0.025 is very close, below 0.050 close, below 0.080 moderate, and beyond that distant.
Closest ancient populations to Parsi
Among ancient genomes, the closest match to the modern Parsi average is Iran LBA Dinkha Tepe (Iranian Plateau Profile) (Late Bronze Age), at 0.061, followed by Iran LBA Hasanlu (Iranian Plateau Profile) and Pakistan High Medieval Ghaznavid Udegram. That is only a moderate match: part of the modern profile was already present then, but later movements of people added substantial layers. A close ancient match is not proof of direct descent: it means those individuals carried a similar overall mix of ancestry.
| # | Ancient sample or group | Period | Distance |
|---|---|---|---|
| 1 | Iran LBA Dinkha Tepe (Iranian Plateau Profile) | Late Bronze Age | 0.0608 |
| 2 | Iran LBA Hasanlu (Iranian Plateau Profile) | Late Bronze Age | 0.0623 |
| 3 | Pakistan High Medieval Ghaznavid Udegram | High Medieval, c. 1000-1300 AD | 0.0637 |
| 4 | Mongolia IA Xiongnu Late (Iranian Plateau Profile) | Iron Age | 0.0639 |
| 5 | Uzbekistan HP Kushan Rabat | Historical period | 0.0693 |
| 6 | Tajikistan HP Kushan Ksirov | Historical period | 0.0715 |
| 7 | Mongolia Early Medieval Uygur Period (South Central Asian Profile) | Early Medieval, c. 500-1000 AD | 0.0716 |
| 8 | Iran IA Hasanlu (Iranian Plateau Profile) | Iron Age | 0.0725 |
| 9 | Xinjiang HP Junmachanyilian (South Asian Profile) | Historical period | 0.0726 |
| 10 | Iran IA Hajji Firuz (Iranian Plateau Profile) | Iron Age | 0.0751 |
Ancient DNA from India
Our ancient DNA database holds 59 individuals excavated in present-day India, dated from about 2,350 BC to 1860 AD. The most frequent Y-DNA haplogroups among them are R2a (4), H1a (3) and J2b (3), and the most frequent mtDNA haplogroups are M3 (6), H1 (4) and U2 (4). These are people who lived on the same land in the past, not necessarily ancestors of today's Parsi population.
Most frequent Y-DNA haplogroups
Ancient DNA studies on India
- Blending borders: reconstructing the genetic history of the Sindhi population (2026)
- Blending borders: reconstructing the genetic history of the Sindhi population (2026)
- The Old Lady spider cave skeletons in Ladakh have diverse maternal genetic origin (2026)
- Layers in the sand: The genetic imprint of migration, culture, and Indus craft in the Thar desert (2026)
- Admixture and Genetic Connectivity: Autosomal Insights Into Indo-Aryan Speakers at the Eastern Edge of the Indian Subcontinent (2026)
- Novel 4400-year-old ancestral component in a tribe speaking a Dravidian language (2025)
- Ancient mitogenomes from Neolithic, megalithic and medieval burials suggest complex genetic history of Kashmir valley, India (2025)
- 50,000 years of evolutionary history of India: Impact on health and disease variation (2025)
Browse them in our ancient DNA database: 59 from India.
Compare yourself with Parsi
Paste your G25 coordinates (scaled, one line, with or without a name in front) and we compute your genetic distance to the Parsi average and to its closest neighbours, right in your browser. Nothing is uploaded or stored.
| # | Population | Your distance | Closeness |
|---|
This list only covers Parsi and its neighbours. To find out which of our reports actually fits your DNA, run the free Report Finder: it runs the same fit test that every report uses before an order.
No G25 coordinates yet? Get a free simulated G25 from your raw DNA file, or order G25 coordinates.
About these numbers
Keep in mind that a population average is a statistical summary of a limited number of sampled individuals. Real people inside any group vary, some carry more of one ancestry and some less, and no genetic profile decides who is or is not Parsi. Use these numbers as a map of deep ancestry, not as a label.
Method: averaged G25 coordinates, Euclidean distances to other modern and ancient averages, and a non-negative least-squares model against deep ancestral reference populations (data generated 2026-10-01). Read the full method.
Go deeper than the average
This page describes the Parsi average. Your own DNA has its own story: Sindhu – South Asian Report (15€) models your genome against the ancient sources and modern communities of the region, era by era, in a personal PDF report.
A report built for people of South Asian ancestry: ancient regional core (Indus Valley Civilization, BMAC, Steppe migration, Himalayan Iron Age & Classical layers) versus non-regional outside admixture, across 13 communities (Punjabi...
More populations from South Asia
Frequently asked questions
What is the ancient genetic make-up of Parsi?
Modelled with deep ancestral reference populations, the Parsi average is about 35.9% Iranian Neolithic farmers, 17.4% South Asian hunter-gatherer related (Irula proxy) and 16.5% Anatolian Neolithic farmers. These proportions are model estimates for a group average, not exact values for any one person.
Which populations are genetically closest to Parsi?
Leaving aside other regional samples listed under the same name in the G25 sheet (such as Parsi (Pakistan)), the closest modern populations to the Parsi average are Qizilbash, Iranian (Persian Khorasan) and Persian (Khorasan). The closest ancient matches are Iran LBA Dinkha Tepe (Iranian Plateau Profile), Iran LBA Hasanlu (Iranian Plateau Profile) and Pakistan High Medieval Ghaznavid Udegram.
Can a DNA test tell me if I am Parsi?
No DNA test can confirm an ethnicity or a nationality. What DNA can show is how similar your genome is to the sampled Parsi average and to its neighbours. If you have G25 coordinates, paste them in the comparison box on this page to check that for free.
Which ExploreYourDNA report suits Parsi ancestry?
Sindhu – South Asian Report is our regional report for this part of the world. On the Parsi average its fit test reads partial, so check your own DNA first: the free Report Finder runs that test on your file.
Where does the data on this page come from?
From the averaged Global25 (G25) coordinates of the sampled individuals: genetic distances to other modern and ancient population averages, and a non-negative least-squares model against deep ancestral reference populations. The full method is described at https://www.exploreyourdna.com/populations#method.