DNA Ethnicity Estimates Explained: Why Your Results Changed (and What They Really Mean)

For many people, the ethnicity estimate is the most exciting part of a DNA test. You open your results and there it is: a colourful map and a set of percentages telling you that you're 38% Irish, 25% English and 12% Norwegian. Then, a year later, you log in and find that your Norwegian has vanished and you've become 9% Swedish.

It's one of the most common questions family historians ask: if my DNA hasn't changed, why have my results? The answer lies in how ethnicity estimates actually work. Once you understand that, the changes make sense, and you'll know how much weight to give the numbers.

What an ethnicity estimate is (and isn't)

An ethnicity estimate is a statistical prediction, not a measurement. It doesn't read a label in your DNA that says "Irish". Instead, it compares segments of your DNA with the DNA of reference groups, people whose families have lived in a particular region for many generations, and asks which groups your DNA looks most like.

Your result is the company's best estimate of which populations your ancestors, perhaps several hundred years ago, most resembled genetically. It's a useful and often fascinating picture, but it's a model, with all the uncertainty that implies.

How the estimate is calculated

1. Reference panels

Each company builds a reference panel: a collection of DNA samples from people with deep roots in specific places. A person whose four grandparents, and ideally earlier generations, all came from western Ireland might be included in an Irish reference group, for example.

2. Comparing your DNA

Your DNA is broken into many small segments. For each segment, an algorithm estimates which reference population it most resembles. Some segments are easy to assign; others look similar to several neighbouring populations.

3. Adding it up

The assignments are combined into percentages, usually with a range showing the uncertainty. That range is important: a result of "12% Norwegian" might really mean "somewhere between 0% and 25%".

Why your results change

Your DNA never changes. What changes is the comparison.

Bigger reference panels

As companies add more people to their reference groups, they can tell neighbouring populations apart more precisely. A region that used to be lumped together may be split into several, or re-drawn entirely.

Better algorithms

Companies periodically improve the methods they use to compare segments. Better methods can reassign segments that were previously misclassified.

New regions

Updates often introduce new, more specific regions. Someone previously shown as broadly "Scandinavian" might now see a split between Norway, Sweden and Denmark, or find that the Scandinavian signal was really part of a broader north-western European picture.

That's why the infamous "disappearing Scandinavian" is so common for people with British ancestry. Centuries of migration around the North Sea mean that British and Scandinavian DNA is genuinely similar, and earlier models sometimes over-assigned British segments to Scandinavia. Better models corrected this.

Why siblings get different results

If you and your brother or sister test, you'll almost certainly get different ethnicity estimates, sometimes noticeably different. This isn't an error.

You inherit about 50% of your DNA from each parent, but which 50% is random. Full siblings share, on average, about half their DNA with each other. So if your mother's ancestry is part English and part Italian, you might have inherited more of her Italian segments than your sister did, and your estimates will reflect that.

That's also why testing a parent can be so revealing: their estimate captures DNA you didn't inherit.

Why different companies give different results

Test with two companies and you'll usually get two different estimates. Each company uses:

• Its own reference panels, built from different people.
• Its own set of regions, with different names and boundaries.
• Its own algorithms.

So one company might call a part of your DNA "Germanic Europe" and another "France" or "Central Europe". Neither is necessarily wrong. They're drawing the map differently.

AncestryDNA, for example, is particularly strong for British, Irish and North American ancestry and offers "communities" that can link you to groups of testers with recent shared origins. MyHeritage is often well regarded for continental European and Jewish ancestry. Our AncestryDNA vs MyHeritage comparison covers the differences.

How accurate are ethnicity estimates?

A fair summary:

• Continental level: generally reliable. If your estimate says you have substantial European, African or East Asian ancestry, that's very likely correct.
• Regional level: often meaningful, but less certain. Distinguishing Ireland from Scotland, or Germany from France, is genuinely difficult because populations have mixed for thousands of years.
• Small percentages: treat anything under about 5% with caution. Small amounts can be "noise", especially if the range includes zero.
• Under-represented populations: results are less detailed for regions and peoples with fewer people in the reference panels, including many Indigenous peoples. A consumer test cannot determine Aboriginal and Torres Strait Islander identity, or any other cultural or community identity.

Communities and genetic groups: the more useful part

Alongside broad ethnicity regions, AncestryDNA's "communities" and MyHeritage's "genetic groups" work differently. Rather than comparing you to ancient reference populations, they identify clusters of testers who share DNA with each other because they descend from people who lived in the same area in the last few hundred years.

For family history, these can be more useful than percentages. Being placed in a community associated with, say, "Cornwall" or "Ulster" suggests you have relatively recent ancestors from that area, which is a concrete lead for research.

How to use your ethnicity estimate sensibly

1. Treat it as a hint, not proof. Use it to suggest where to look in the records, not to settle questions about your family tree.
2. Look at the ranges. A small percentage with a range including zero may not be real.
3. Compare with your documented tree. If your tree says you have Italian grandparents and your estimate shows Italian ancestry, that's reassuring. If there's a big surprise, look at your DNA matches before jumping to conclusions.
4. Focus on matches for family history. Your list of DNA relatives is far more useful than your ethnicity for building a family tree. Our guide to what to do when your results arrive explains how to work with matches.
5. Expect updates. Your estimate will probably change again. Note down your current results if you want to compare them later.

When ethnicity results are genuinely surprising

Sometimes an estimate reveals something unexpected: a large percentage of ancestry from a region nobody in the family has mentioned, or the complete absence of ancestry you expected. Before drawing conclusions, consider:

• Is the result within the range of normal variation, or is it a significant amount, such as 25% or more?
• Could the "missing" ancestry be grouped under a different region name?
• Do your DNA matches support the surprise? Matches with close relatives from an unexpected background are much stronger evidence than percentages alone.

If your matches suggest an unknown parent or grandparent, take your time and look after yourself. These discoveries can be emotionally significant. Organisations such as DNA-focused support groups and counselling services can help, and there's no obligation to contact anyone until you're ready.

How far back does an ethnicity estimate reach?

This is a common source of confusion. Ethnicity estimates don't refer to a specific generation. Instead, they reflect the populations your ancestors belonged to over many generations, sometimes stretching back several hundred years or more.

That's one reason an estimate can seem to contradict a well-documented family tree. Suppose your tree shows that all eight of your great-grandparents were born in England. Your estimate might still show some Scottish, Irish, Welsh or even Scandinavian ancestry, because people moved around Britain and northern Europe for centuries before your documented ancestors were born. The estimate isn't wrong; it's simply looking further back than your tree.

Equally, you may have a documented ancestor from a particular country but little or none of that ancestry in your estimate. Within five or six generations, it's entirely possible to inherit very little DNA from a particular ancestor, and they may have carried DNA typical of a neighbouring region.

Ethnicity versus your DNA matches

If you're trying to answer a specific family question, such as whether a great-grandfather really came from Italy, your matches will usually give a clearer answer than your ethnicity. Matches with people who still live in Italy, or whose documented families came from the same region, are concrete evidence of a connection. Ethnicity percentages are broad brushstrokes; matches are fingerprints.

Learn how to work with matches in our guides to centimorgans and your first steps after results arrive.

Which test gives the best ethnicity estimate?

There's no universal winner, because it depends on where your ancestors came from. For most Australians and others with British, Irish and North American roots, AncestryDNA's combination of detailed regions and communities is hard to beat, and its huge database gives you the most relatives too.

For continental European and Jewish heritage, MyHeritage DNA's genetic groups are often especially informative.

Our 60-second quiz can help you decide.

A quick checklist for reading your estimate

• Is the region at the continental level, such as "Europe", or a more specific region? Continental results are the most dependable.
• What is the range? If it includes zero, the region may not be present at all.
• Has the company updated your estimate recently? Note the date, so you can compare future versions.
• Do you belong to any communities or genetic groups? These often point to more recent and more specific origins.
• Does the estimate fit your documented tree, allowing for the fact that it reaches further back in time?
• If it doesn't fit, do your DNA matches support the surprise?

Working through these questions turns a colourful chart into genuinely useful evidence, and protects you from reading too much into any single number. Above all, remember that your ethnicity estimate is just one part of your results. The relatives on your match list are where the real family history discoveries are made.

Frequently asked questions

Why did my Scandinavian ancestry disappear?

Probably because an update improved the company's ability to tell British and Scandinavian DNA apart. Centuries of migration mean these populations share a lot of DNA, and older models sometimes over-assigned British segments to Scandinavia.

Can I see my old ethnicity results?

Some companies let you view previous versions for a time after an update. Others don't, so it's worth saving a screenshot of your results.

Why don't I have any DNA from an ancestor I know about?

You inherit a random half of each parent's DNA, so DNA from more distant ancestors can be diluted or lost entirely within a few generations. Beyond about five or six generations, you may share little or no DNA with some genuine ancestors.

Is my ethnicity estimate the same as my nationality?

No. Ethnicity estimates relate to genetic similarity with reference populations, not citizenship, culture or identity.