If you have shopped for a paternity test recently, you have probably seen a phrase like "23 autosomal STR markers" or "tests more than 20 loci" buried in the product details. The words sound technical, and most product pages do a poor job of explaining what any of it means. That is a shame, because the concept is not complicated once someone walks you through it, and understanding it makes the reported probability numbers on your results a lot less mysterious.
This article is a plain-language explainer for the person deciding whether to order a home paternity test. No genetics degree required. By the end you will know what "autosomal" refers to, what an STR is, why paternity labs use these particular markers instead of the more famous Y-chromosome or mitochondrial tests, and roughly how the math produces a number like 99.99%.
What "autosomal" actually means
Humans have 46 chromosomes arranged in 23 pairs. Twenty-two of those pairs are called autosomes. The final pair is the sex chromosomes: XX in most females, XY in most males. That is the whole vocabulary. "Autosomal" simply means "on one of the 22 non-sex-chromosome pairs."
Why does that distinction matter? Because autosomes are inherited the same way regardless of the child's sex. You get one copy of each autosome from your biological mother and one from your biological father, and the same is true for every human. A boy and a girl inherit their autosomes by the identical mechanism. That symmetry is exactly what a paternity test needs. It works whether the child is a son or a daughter, and it works for testing between fathers, mothers, siblings, grandparents, aunts, and uncles without swapping methodology in the middle. You can read a short definition of autosome, chromosome, and other genetics terms at the NHGRI Talking Glossary.
Sex chromosomes are a different animal. A father passes his Y to his sons but his X to his daughters. Mothers pass an X to everyone. That asymmetry is useful for some questions (which we will get to), but it makes the sex chromosomes a poor tool for standard paternity work.
What an STR is
STR stands for short tandem repeat. Picture a stretch of DNA that looks like this:
...AGAT AGAT AGAT AGAT AGAT AGAT AGAT AGAT...
The four-letter sequence AGAT is a repeat unit. It is short (usually 2 to 6 base pairs long), and it appears many times in a row, or "in tandem." Different people carry different numbers of repeats at the same physical location in their genome. One person might have 11 AGAT repeats at a particular spot; another person might have 14. That count is the STR's allele.
Because you inherit one copy of each autosome from each parent, at any given STR location you carry two alleles: one from mom, one from dad. Your genotype at that location might be written 11,14, meaning one chromosome has 11 repeats and the other has 14. A child's genotype at a marker must be explainable by the alleles the biological parents actually carry. If a child shows 11,14 and the tested man carries 8,9, he cannot have contributed the 11 or the 14 to that marker. That is the whole logical core of a paternity test.
The National Institute of Standards and Technology maintains a public reference database of the STR loci used in forensic and relationship testing, including their repeat structures and known allele frequencies. If you want to see the raw specifications, the NIST STR DNA Database is the authoritative public source.
Why labs pick autosomal STRs instead of Y or mitochondrial DNA
There are three broad ways to compare human DNA for relationship questions, and only one of them answers the paternity question cleanly.
Y-chromosome STRs are inherited from father to son with almost no shuffling. That sounds ideal for paternity, but it has a hidden flaw. Every man in the same paternal line carries essentially the same Y-STR profile. A tested man's brother, father, paternal uncle, and paternal grandfather will match a boy the same way the biological father would. Y-STRs cannot distinguish between a father and the father's brother, so they cannot answer "is this specific man the biological father" on their own. They are also useless for testing a daughter, because daughters do not carry a Y chromosome.
Mitochondrial DNA is inherited from the mother only, and it changes slowly across generations. It is a powerful tool for tracing a maternal line back many generations. It is a poor tool for paternity for the mirror-image reason Y-STRs are: it says nothing at all about the father, and everyone in the same maternal line shares essentially the same mtDNA sequence.
Autosomal STRs avoid both problems. Each parent contributes one allele at each marker, the choice of which allele is essentially random, and the markers used sit on different chromosomes so they behave as independent events. That last part is what lets the statistical math work. When labs run enough independent autosomal markers, the probability that an unrelated man happens to share every allele the child needs from the father becomes vanishingly small.
How many markers labs actually run
The industry has settled on a small set of well-characterized autosomal STR loci that have been studied for decades. The FBI's Combined DNA Index System, which most people know as CODIS, defines a core set of 20 loci used in United States forensic casework. You can read the FBI's overview of that system at the FBI CODIS page. Those same loci form the backbone of modern paternity testing.
A typical breakdown looks like this:
- 20 CODIS core loci — the baseline that most reputable relationship labs cover.
- 23 autosomal STR markers — the higher-tier standard now common in home paternity products, adding a few markers beyond the CODIS 20 for extra discriminating power. This is the standard US Diagnostics Center panel.
- Up to 28 total markers — a reserve of additional autosomal loci that a lab can bring online for difficult cases, typically kinship questions where a first-degree relative is not available.
Running more markers on a straightforward duo (child and alleged father) rarely changes the answer. The 23 already push probabilities into the 99.99%+ range for a true match. The reserve markers earn their keep on indirect relationship questions, such as grandparent, sibling, or aunt/uncle tests, where the biological signal is weaker because there are more generations of dilution between the tested parties. The International Society for Forensic Genetics publishes guidelines on how to handle these harder cases, and their ISFG recommendations shape how labs across the world approach inconclusive kinship results.
How the math works — one marker, then many
Take a single autosomal STR, say the marker called D3S1358 for the sake of example. Suppose the child's genotype at that marker is 15,17, the mother's is 15,16, and the tested man's is 14,17.
The child needed a 15 or a 16 from mom. She has both — the 15 is a match to mom. That means the 17 in the child had to come from the biological father. The tested man carries a 17. So far, he is consistent with being the father at this marker. That single observation is not proof; plenty of unrelated men also carry a 17. What matters is how common the 17 allele is in the general population. If the 17 is present in, say, 25% of unrelated men, then this marker's contribution to the "he is the father" side of the ratio is 1 divided by 0.25, which equals 4. The math is done relative to a reference population, which is why labs report which population database they used.
Now do that at 23 different markers. Each marker sits on a different chromosome, so the alleles at each marker are inherited independently. That means the individual ratios multiply. If every marker contributed a factor of 4 (they will not, but bear with the round number), 23 markers would produce a combined ratio of 4^23, or about 70 trillion. In practice the individual factors vary, and the final combined likelihood ratio for a true father typically lands somewhere between one trillion and one quadrillion. That is then converted into the probability of paternity you see on your report, which is usually written as 99.99% or higher.
The reason you never see 100% on a paternity test is that no finite population database can rule out the theoretical possibility of an untested identical twin. Everything short of that gets absorbed into the 99.99%.
What autosomal testing does NOT do — and when labs use Y or mtDNA anyway
Autosomal STRs are the right tool for standard paternity, maternity, and most kinship questions. They are not the right tool for every DNA question a family might have.
If two men want to know whether they share a paternal line — for example, half-brothers through the same father, where neither the father nor a common paternal relative is available for testing — a Y-STR comparison can answer that specific question when autosomal markers cannot deliver a conclusive result. It will not identify a single man; it will only confirm or exclude the shared paternal line.
If a group of people wants to know whether they share a maternal line — grandmother, mother, daughters, granddaughters through the female line — mitochondrial DNA analysis can do that. Again, it will not distinguish between individual women in that line, only confirm or exclude the line itself.
These are specialized reflex tests that a lab may recommend after an autosomal panel returns an inconclusive kinship result. For a routine home paternity test ($79), you should not need them. The autosomal panel handles the question directly.
How USDC's up to 28 marker workflow uses this
At US Diagnostics Center, every relationship test starts on the same 23 autosomal STR foundation. For a straightforward alleged-father-and-child comparison, those 23 markers produce a definitive result the vast majority of the time. That is where the process ends for most paternity cases.
The workflow changes for harder questions. On a home grandparent test ($139), a home sibling test ($139), or a home aunt/uncle test ($139), the biological signal is thinner by design — grandparents share about 25% of their DNA with a grandchild, half-siblings around 25%, aunts and uncles around 25%. The 23 core markers will still give a clear answer most of the time. When they do not, USDC's protocol is to reflex-run additional autosomal markers, bringing the total up to 28, before returning an inconclusive verdict. Those extra markers are held in reserve specifically for this purpose. They do not fire on paternity trios; they are reserved for kinship inconclusives.
This is the practical reason USDC's marker phrase is "up to 28 genetic markers." The 23 do the work on most cases. The reserve exists so that a harder kinship question does not get closed out prematurely.
If your situation involves a legal chain-of-custody requirement rather than a private home test, a legal paternity test uses the same autosomal STR science with witnessed collection. That product is coming soon.
Frequently asked questions
Are all 23 markers on different chromosomes?
Most of them are, and the ones on the same chromosome sit far enough apart to be inherited independently in practice. That independence is what lets the individual marker probabilities multiply together. If two markers were tightly linked, the math would double-count them and the reported probability would be inflated. Established relationship-testing panels are designed specifically to avoid that problem.
Do more markers always mean a more accurate test?
Up to a point. On a straightforward paternity trio, the improvement from 20 markers to 23 markers is small, and the improvement from 23 to 28 is smaller still — the answer was already at 99.99%+ with 20. Extra markers matter most on indirect relationship tests (grandparent, sibling, aunt/uncle) where the base signal is weaker. That is why USDC only runs the additional markers on kinship reflex, not on every paternity case.
Why isn't the result ever exactly 100%?
Because the calculation compares the tested man to a population database, and no database can formally rule out an untested identical twin of the true father. In practice, a probability of paternity of 99.99% or higher is the standard threshold for a positive result.
Does the same STR panel work for maternity tests too?
Yes. Because autosomal inheritance is symmetric, the exact same 23-marker approach works for a home paternity test or a home maternity test. The lab is just checking which alleles the child inherited from each parent. Maternity is priced at $129.
Can I request a Y-STR or mitochondrial test if my autosomal result is inconclusive?
Sometimes, depending on the biological question. Y-STRs are only informative for a purely paternal line, and mtDNA only for a purely maternal line. If your first result is inconclusive, the right next step depends on who is available to test and what specific relationship you are trying to confirm. A lab that runs both autosomal and lineage-marker workflows can guide you through the options.
0 comments