How Accurate Are DNA Tests for Ethnicity? | What The Numbers Show

Most people want to know how accurate DNA tests for ethnicity really are, and the honest answer depends on what level of detail you need. For broad continental origins—European, African, East Asian—the estimates are reasonably reliable, with accuracy typically in the 70 to 95 percent range for well-represented populations. But those precise-looking numbers for individual countries or regions are statistical clues, not definitive facts, and they can shift every time the company updates its database or refines its algorithm. Understanding what these numbers actually represent helps you avoid reading too much into a single percentage.

What Does A DNA Test Actually Tell You About Your Ancestry?

Consumer DNA tests analyze roughly 700,000 genetic markers from a saliva sample and compare them against the company’s reference panels—groups of people whose DNA has already been linked to specific regions. AncestryDNA examines over 700,000 locations and 23andMe tests around 690,000, with both advertising better than 99 percent accuracy for each individual marker. That raw-data accuracy, however, refers to whether the machine read the DNA letter correctly, not to the ethnicity estimate itself. This distinction between raw-data accuracy and estimate accuracy is the single most misunderstood aspect of consumer DNA testing.

The ethnicity estimate is a statistical inference: “your DNA pattern resembles people who currently live in this region.” Its reliability depends entirely on whether your ancestral populations are well-represented in the reference panel. Regions with dense sampling produce tighter estimates; underrepresented populations and historic mixing produce broader, less stable ones. One published benchmark found that a particular ancestry algorithm averaged 84.2 percent accuracy overall, but performance dropped as low as 32 percent for some subgroups. Academic researchers have confirmed that ancestry inference is inherently comparative, not absolute—your results reflect who is in the reference dataset, not every population that exists.

Why The Same Person Gets Different Results From Different Companies

AncestryDNA and 23andMe maintain separate reference databases and use different statistical algorithms, so the same DNA sample can produce noticeably different percentage estimates from each service. A person with mixed European and Indigenous American ancestry might receive 75 percent European from one company and 68 percent from the other—both estimates can be reasonable within their own systems.

These differences don’t mean either company is wrong. They reflect real variation in how ancestry inference works: the reference populations, the markers tested, and the statistical method all shape the output. As Wired’s deep look at ethnicity estimate mechanics explains, the numbers are best understood as population-level probabilities rather than personal certainties.

Three Traps That Skew How People Read Their Results

1. Treating percentages as fixed facts. Results frequently shift after a database update or algorithm refresh—not because your DNA changed, but because the company improved its reference panel. A 5 percent change in a single region is common after an update.

2. Mistaking country labels for ancient borders. DNA doesn’t recognize modern political boundaries. A label like “Germany” on a report means “central Europe with a strong match to German reference samples,” not that a specific ancestor lived in modern Germany.

3. Trusting small percentages. Minority admixture estimates below 5–10 percent are the least stable part of any report. They often appear, disappear, or move between regions when the database updates, especially for populations with sparse reference sampling.

If you are ready to explore your own background, our comparison of the best ethnicity DNA tests breaks down which service suits different ancestry goals and budgets.

The most useful way to read an ethnicity estimate is as a genealogical conversation starter, not a definitive statement of identity. Broad continental regions are reasonably trustworthy; fine-grained country numbers are hints worth investigating further. When a result surprises you, check whether the company has released a recent database update before drawing conclusions—and remember that no test can map your ancestors’ exact homeland, only where similar DNA is found today. Use the broad picture for curiosity and the specific numbers as breadcrumbs for family research.

FAQs

Can my ethnicity percentages change over time?

Yes—companies regularly expand their reference panels and refine their algorithms, which can shift your reported percentages. A change of 5 percent or more in a specific region after a major update is not unusual and does not mean the earlier result was wrong, only that the reference data has improved.

Is one company more accurate than another?

No single service is universally “most accurate” because reliability depends on your specific genetic background. A company with a larger reference panel for your ancestral populations will generally produce more stable estimates for you, but results still vary across platforms. Checking each company’s reference regions before buying can help you choose.

Do these tests work well for people with mixed ancestry?

They work but with broader and less stable estimates. People from underrepresented populations or with highly mixed backgrounds often receive wider percentage ranges, and small contributions from specific populations may not register at all. The broad continental picture is usually reliable, but fine-grained country-level detail is less dependable.