Siblings can absolutely get different ancestry results from the same DNA testing company, and the differences are real, not errors. Each child inherits a unique combination of DNA from the same two parents, which means one sibling might show 25% of a given ancestry and another might show 15%. The biology behind this is straightforward, but the way testing companies interpret DNA adds another layer of variation that can make siblings’ results look even more different than their actual genetic inheritance would suggest.
Why Siblings Don’t Share Identical DNA
You and your siblings each got roughly half your DNA from your mother and half from your father, but the specific halves are different every time. Before an egg or sperm cell is created, the parent’s chromosomes shuffle segments in a process called recombination. This means each reproductive cell carries a unique patchwork of DNA from that parent’s own two parents (your grandparents). Because the shuffling happens independently each time, two siblings end up with overlapping but non-identical slices of their parents’ genomes.
Full siblings share about 50% of their DNA on average, but the actual number varies. Some sibling pairs share closer to 40%, others closer to 60%. The differences in which specific segments each sibling inherits are what drive different ancestry percentages. If your father has, say, both West African and Northern European ancestry in his genome, one child might inherit more of the segments tracing back to West Africa while another inherits more of the segments tracing back to Northern Europe. Neither result is wrong. They just reflect different draws from the same genetic deck.
Research on sibling genetic variation confirms that these differences stem directly from the distinct recombination patterns that occur during sexual reproduction.
The Difference Between Your Family Tree and Your DNA
Here is something that surprises most people: you have ancestors who contributed nothing to your DNA. Go back enough generations and the math makes this inevitable. You have four grandparents, eight great-grandparents, sixteen great-great-grandparents, and so on. By the time you reach ten generations back, you have over a thousand positions in your family tree. But your DNA can only be divided into so many segments before those segments become vanishingly small. Eventually, some ancestors’ contributions get lost entirely through the randomness of inheritance.
This means that not all of a person’s genealogical ancestors are also their genetic ancestors, and some ancestors simply won’t be reflected in their DNA at all.1Ancestry Reimagined. Genealogical and Genetic Ancestry Two siblings might share a great-great-grandparent on paper, but only one of them might have actually inherited any DNA from that particular ancestor. If that ancestor came from a specific region, only the sibling who inherited their DNA will show that region in their results.
This is why ancestry results can diverge more dramatically for the smaller-percentage ancestries. If 3% of a parent’s DNA traces to a particular region, one child might inherit that chunk and show 2% in their results, while another child might not inherit it at all and show 0%. For the majority ancestries, the differences between siblings tend to be more modest, usually just a few percentage points in either direction. But for trace ancestries, one sibling might see a region appear and the other might see it completely absent.
How Much Comes from Biology and How Much from the Algorithm
The variation between siblings isn’t just about which DNA segments they inherited. A significant part of it comes from how the testing company’s software interprets those segments. Ancestry tests work by comparing your DNA to reference panels, which are databases of DNA from people with well-documented ancestry from specific regions. The algorithm looks at stretches of your genome and asks which reference population each stretch most closely resembles, then reports the estimated proportions.
Different companies use different reference panels, different statistical methods, and different ways of grouping populations. This means the same person tested by two different companies can get meaningfully different results. A study of identical twins, who share virtually 100% of their DNA, demonstrates this clearly. When identical twin pairs were tested by the same company, their ancestry results matched closely, with agreement rates averaging between about 95% and 99%. But when the same twins were tested by different companies, agreement dropped substantially, averaging between roughly 53% and 84%.2PubMed Central. Consistency of Direct to Consumer Genetic Testing Results Among Identical Twins – Section: Results
Think about what that means. Identical twins have the same DNA, so any difference between their results is entirely the algorithm’s doing. If a single company can produce 95–99% agreement for identical twins but a cross-company comparison drops to 53–84%, the software’s interpretation is contributing a lot of the perceived variation. Now imagine two non-identical siblings who share only about half their DNA. They are experiencing both the biological variation (different DNA segments) and the algorithmic variation (how those segments are categorized) stacked on top of each other.
Why Small Percentages Are Especially Unreliable
If you and your sibling are both showing, say, 45% British and Irish ancestry, those numbers will probably be within a few points of each other. The larger an ancestry component is, the more DNA segments contribute to the estimate and the more stable it becomes. But once you get down to single-digit percentages, the estimates become much noisier.
A 2% result for a given region might be based on just a few short stretches of DNA. Whether the algorithm assigns those stretches to Region A or the closely related Region B can swing a small percentage by several points or make it appear and disappear entirely. This is partly a biological effect, since a sibling may simply not have inherited those few segments, and partly an algorithmic one, since the software’s confidence in assigning short DNA stretches to a specific population is inherently lower.
Companies typically acknowledge this in their confidence-level settings. Most ancestry platforms let you toggle between a more speculative estimate and a more conservative one. At the conservative setting, the software only reports ancestry it is fairly confident about, which often means small percentages vanish and get absorbed into a broader “unassigned” or neighboring category. If you and your sibling are comparing results, you should both be looking at the same confidence level, because switching between the two can change the picture dramatically even for a single person.
What the Algorithms Are Actually Doing Under the Hood
Ancestry estimation is a problem called local ancestry inference: the software walks along each of your chromosomes, segment by segment, and tries to figure out which ancestral population each segment came from. The accuracy of this process depends on several factors, including how well the DNA data has been organized computationally and how long ago the ancestral mixing occurred.
Recent research comparing different approaches to local ancestry inference has found that methods working with raw, unprocessed genetic data can produce results with roughly 3–5% higher accuracy than methods that first require an additional processing step called phasing.3PubMed Central. Phase-free local ancestry inference mitigates the impact of switch errors on phase-based methods That might sound like a technical detail, but it matters because it means two companies using different computational approaches could assign the same DNA segment to different populations. When this happens across thousands of segments, the reported percentages shift.
Another factor is how recently the ancestral mixing occurred. If your family’s admixture happened many generations ago, the segments tracing to each ancestral group have been broken up into smaller and smaller pieces by recombination over time. Shorter segments are harder for algorithms to assign confidently. This is one reason why recent admixture (within the last few generations) tends to show up more consistently between siblings than ancient admixture.
Reference Panels and Regional Boundaries
The reference panels that companies use to categorize your DNA are not fixed truths about human genetics. They are samples from real people living in specific places today, and the way those samples are grouped into labeled populations involves human judgment. Where one company draws the line between “French” and “German” ancestry, or between “West African” and “Central African” ancestry, depends on their reference data and their statistical grouping decisions.
These reference panels also evolve. Companies periodically update their databases as they collect more DNA samples, which can change your results without any change to your actual DNA. If you tested two years ago and your sibling tested last month, you might be comparing results generated by different versions of the algorithm using different reference data. Most companies will re-run your results against their updated panels, but you’d need to check whether both sets of results reflect the same version.
Populations that are genetically similar to each other are especially prone to being reshuffled between updates. Scandinavian and British ancestry, for instance, can be difficult for algorithms to distinguish cleanly because of extensive historical mixing between those populations. One update might assign a segment to “Scandinavian” that a previous version assigned to “British and Irish.” Neither assignment is exactly wrong; the DNA genuinely has affinities to both. The algorithm just has to pick one, and different statistical approaches pick differently.
When Sibling Differences Should Raise Questions
Most of the time, differing ancestry results between siblings are completely expected and harmless. But occasionally the differences are large enough to suggest something beyond the normal range of variation. If two full siblings share only about 25% of their DNA instead of the expected 50%, they may be half-siblings rather than full siblings, meaning they share one biological parent but not both. Ancestry companies report estimated DNA sharing between matched relatives, and that number is more informative than the ancestry percentages themselves.
A full sibling pair will typically share between roughly 33% and 50% or so of their DNA, with the average near 50%. Half-siblings share about half that. If one sibling shows a large ancestry component that the other completely lacks and it isn’t just a trace amount, it might be worth looking at the DNA-sharing percentage to understand the relationship. Some families have discovered through consumer testing that siblings have different biological fathers, which obviously produces much larger ancestry differences than recombination alone could explain.
That said, jumping to conclusions based solely on ancestry percentages is a mistake. Two full siblings can easily differ by 10 or more percentage points in a given category through normal inheritance variation alone, especially for mid-range ancestries. The DNA-sharing metric, not the ancestry pie chart, is the reliable indicator of biological relationship.
Comparing Results Across Different Companies
If you tested with one company and your sibling tested with another, you should expect substantial differences in the ancestry breakdown even before accounting for the biological variation between you. As the identical-twin study showed, cross-company agreement for people with the exact same DNA can be as low as 53%.2PubMed Central. Consistency of Direct to Consumer Genetic Testing Results Among Identical Twins – Section: Results For siblings who already differ biologically, cross-company comparisons are nearly meaningless for fine-grained percentages.
If you genuinely want to compare ancestry results between siblings, test through the same company during the same time period so you’re comparing results generated by the same algorithm and the same reference panel. Even then, expect differences. The biology guarantees them. But at least you’ll be comparing apples to apples rather than apples to the output of a completely different statistical model.
One practical upside of testing multiple siblings through the same company is that it can collectively reveal more about your parents’ ancestry than any single test. Since each sibling inherited different segments, one sibling might have inherited the DNA that reveals a particular ancestral component that another sibling missed entirely. Pooling the information from several siblings gives a more complete picture of the parents’ genomes, even without testing the parents directly.
Ancestry You Won’t Find in DNA
There is a common expectation that an ancestry test will confirm every branch of your known family tree. It won’t, and this is not a flaw in the test. As noted earlier, the random nature of inheritance means DNA from more distant ancestors can simply fail to be passed down to you.1Ancestry Reimagined. Genealogical and Genetic Ancestry If your great-great-grandmother was from a particular ethnic group, there’s a real chance you carry no detectable DNA from her at all. Your sibling, however, might.
This becomes more pronounced with each passing generation. For ancestors within the last three or four generations, you are very likely to have inherited at least some of their DNA. Beyond that, the probability drops. By about seven or eight generations back, it becomes likely that some of your ancestors left no genetic trace in you whatsoever. This doesn’t mean they weren’t your ancestors; it means the biological lottery didn’t preserve their contribution in your particular genome.
For people with known mixed heritage several generations back, this has real implications. An Indigenous American or African ancestor five or six generations back might appear in one sibling’s results and not another’s, or might not appear in either sibling’s results despite solid genealogical documentation. The absence of a given ancestry in a DNA test does not disprove family oral history or documentary records. It just means the DNA pathway happened not to carry that particular signal forward to you.
How Endogamy Complicates the Picture
Endogamy refers to the practice of marrying within a relatively closed community over many generations. Groups such as Ashkenazi Jewish populations, some island populations, and various historically isolated communities show a distinctive genetic pattern: members of the community share unusually long and numerous DNA segments with each other, because their ancestors overlapped extensively.
For siblings from endogamous backgrounds, ancestry results can be both more similar and more confusingly specific. The siblings will usually agree on the broad community-level ancestry (both will show a high percentage for that group), but the way the algorithm handles the closely shared DNA within the community can produce odd small differences in sub-regional estimates. Testing companies have gotten better at recognizing endogamous patterns, but it remains an area where the algorithms can produce quirky results, especially for sub-regional breakdowns within the endogamous group.
Endogamy also affects the DNA-sharing estimates between relatives. Siblings from endogamous backgrounds may appear to share slightly more DNA than typical siblings, because some of their shared segments come not just from their immediate parents but from the deeper overlap in their family tree. This doesn’t change their actual biological relationship, but it can make the relationship-prediction tools slightly less precise.
What Updates and Reprocessing Can Change
If you and your sibling both tested through the same company but at different times, your results may have been generated against different reference panels. Most major testing companies have issued multiple updates to their ancestry algorithms since launching. Each update typically refines regional boundaries, adds new reference populations, and can shift percentages by meaningful amounts.
A sibling who tested in 2019 and checks their results today will likely see numbers that differ from what they were originally shown, because the company has reprocessed the raw data against a newer model. If you’re comparing your results from a recent test to your sibling’s original results from several years ago, make sure both of you are looking at the most current version. Most companies notify users when updates are available and allow reprocessing at no extra charge.
These updates generally improve accuracy, particularly for underrepresented populations. Early reference panels were heavily skewed toward European samples, which meant ancestry from African, Asian, Indigenous American, and Oceanian populations was often poorly resolved or lumped into broad categories. As reference databases have grown, estimates for these populations have become more detailed, which can cause significant shifts in reported percentages between an older and a newer version of the same company’s analysis.