You ordered a paternity test expecting a range of possible answers. What you got was a number that practically screamed certainty: 99.9999% probability of paternity, or a Combined Relationship Index so low it rounds to zero. If you were expecting something in the middle, you are not alone. Almost everyone expects DNA results to look like a school exam, where scores scatter across a spectrum. The reality is the opposite, and understanding why reveals something genuinely surprising about the mathematics of genetic identity.
What STR Testing Actually Measures
Modern paternity testing relies on Short Tandem Repeats, or STRs. An STR is a short sequence of DNA bases that repeats end-to-end at a specific location in the genome called a locus. The number of repeats at any given locus varies from person to person. At one locus, you might carry 11 repeats on the chromosome inherited from your mother and 14 repeats on the chromosome inherited from your father. That two-number combination is called your genotype at that locus.
A laboratory compares the child's genotype at each locus to the alleged father's. If the alleged father is the biological father, the child must have inherited one of his two alleles at every single locus. If even one locus shows a child allele that is impossible to have come from the tested man, he is excluded as the biological father. Full stop.
USDC's laboratory analyzes up to 28 genetic markers, giving the analysis exceptional resolution. The standard panel covers 23 autosomal STR markers. For cases where the initial result lands in an inconclusive range, the lab can extend the analysis with additional markers, up to that 28-marker ceiling.
The Multiplication Rule and Why It Eliminates the Middle
Here is the core of the bimodal mystery. Each STR locus is statistically independent of every other locus. Independence means you can multiply the probabilities together, just as you multiply the odds of flipping heads twice in a row (0.5 times 0.5 equals 0.25).
Suppose a random, unrelated man has a 10% chance of matching the child at a single locus purely by coincidence. That sounds plausible. But multiply that across 23 independent loci, and his probability of matching at all of them simultaneously is 0.10 raised to the 23rd power: a number so small it has 22 zeros after the decimal point before you reach a non-zero digit.
Now consider the true biological father. He matches the child at every locus not by coincidence but by heredity. His Paternity Index at each locus is calculated relative to a reference population. When you multiply those individual indices together, the Combined Paternity Index (CPI) balloons into the millions, billions, or higher. Converting that CPI into a probability percentage using Bayes' theorem, with equal prior odds, produces a figure that essentially pegs the meter at the high end.
The result is two sharp spikes in the probability distribution: one near zero for unrelated men, one near 100% for biological fathers. No bell curve. No crowded middle. This is the bimodal nature of STR paternity testing, and it is not a lab artifact. It is a mathematical inevitability.
Order your Home Paternity Test →
Understanding CPI, PI, and the Probability of Paternity
Labs report two related numbers. The first is the Paternity Index, or PI, calculated locus by locus. The PI at a given locus answers this question: how much more likely is it that the child inherited this allele from the alleged father than from a random man drawn from the same population? A PI of 500 at one locus means the alleged father is 500 times more likely to be the source than a random man.
Multiply all the individual PIs together and you get the Combined Paternity Index, or CPI. The CPI is then converted to a Probability of Paternity using Bayes' theorem with a prior probability of 0.5 (a neutral starting assumption that neither favors nor disfavors paternity before the DNA is examined). A CPI of 10,000, for example, converts to a Probability of Paternity of approximately 99.99%.
Because the CPI compounds multiplicatively across independent loci, even a modest PI at each individual locus snowballs into an enormous CPI for a true biological father. For an unrelated man, even a few mismatched loci drive the CPI to zero instantly, since any locus with a PI of zero nullifies the entire product.
For a deeper look at the statistical frameworks that govern forensic and kinship DNA analysis, the National Institute of Standards and Technology (NIST) forensic DNA program publishes guidelines that laboratories use to validate their STR panels and population databases.
What Happens in the Rare Inconclusive Zone
If the math pushes results so strongly toward the extremes, why does any result ever land in the middle? The honest answer is: rarely, and usually for specific reasons.
Mutations
At a small fraction of loci, a child may carry an allele that differs from both of a parent's alleles by exactly one repeat unit. This is a germline mutation, not evidence of non-paternity. Mutation rates for STR loci are well-documented. When a lab sees a one-step discrepancy at a single locus while every other locus matches perfectly, the resulting PI at that locus is not zero but a small positive number that accounts for the known mutation rate. The CPI remains strongly positive, though lower than it would be without the discrepancy.
Close Biological Relatives
Identical twins share the same STR profile. Full brothers share, on average, half their alleles. When an alleged father is a close biological relative of the actual father, the STR comparison will show more matching than a random man would, pulling the probability upward from near-zero without reaching the typical 99.99%+ range. This is a known limitation the lab will flag in the report.
The Inconclusive Range and Extended Panels
USDC's lab defines an inconclusive Combined Relationship Index (CRI) as one falling between 0.1 and 10. When a result lands there, the laboratory can extend the analysis to bring the total markers tested up to 28, narrowing the statistical uncertainty. Most extended analyses resolve the question definitively in one direction.
The NIH-published research on STR population databases explains how allele frequency data from diverse populations is used to calculate accurate PI values, which is a critical factor in keeping inconclusive results rare.
The Underlying Science Is the Same as Court-Admissible Testing
A common misconception is that at-home paternity kits use a simpler or less accurate science than legal testing. That is not true. The underlying lab science is the same as court-admissible testing. Both use STR analysis, the same population databases, and the same probability calculations. The difference between a home-use test and a legal test is not the science. It is the sample collection process.
Court-admissible paternity testing requires witnessed specimen collection and a documented chain of custody so that a court can be confident the samples belong to the individuals named. Because home kits involve self-collected swabs without independent witnessing, they are designed for personal knowledge only. If you need results for a legal proceeding such as a custody case, child support determination, or immigration matter, a separate legal-use collection process is required. You can learn more about that distinction in our related article on the difference between legal and home paternity tests.
Timelines: When Will You See That Number?
Once your swabs arrive at the laboratory, standard processing takes 2 to 3 business days for the core analysis. Total turnaround from kit order to online result is typically 7 to 10 business days under standard conditions, accounting for kit shipping, lab intake, analysis, quality review, and secure result delivery. If the first-pass result lands in the inconclusive range and additional markers are tested, that may add time to the analysis phase. Expedited result options are available during checkout for those who need answers sooner.
To understand what your result report contains and how to read the numbers, see our guide on how to read your DNA paternity test results.
Why "Probably the Father" Is Not a Real Category
People sometimes ask whether a Probability of Paternity in the 70% or 80% range means the man is "probably" the father. In practice, STR testing almost never produces such a number for a genuinely ambiguous case. A result in that range would almost always indicate one of the specific scenarios described earlier: a mutation, a close biological relative as the actual father, or a sample quality issue. It is not a reflection of biological ambiguity between two unrelated men.
Put simply: if the man is biologically the father, the math says so with overwhelming certainty. If he is not, the math says that too. The distribution of results has two peaks, not a broad hump, and that is precisely what makes STR paternity testing the gold standard for biological parentage determination.
The AABB (formerly the American Association of Blood Banks) sets voluntary accreditation standards for relationship DNA testing laboratories and publishes guidance on reporting requirements, including minimum thresholds for inclusion and exclusion conclusions, that reflect this bimodal statistical reality.
Browse All DNA Tests →
Frequently Asked Questions
Why do paternity test results almost always come back above 99% or below 1%?
Because STR markers are statistically independent, the probability of a match at each locus multiplies together. For a true biological father, these multiplied probabilities produce a Combined Paternity Index in the millions or higher, translating to a Probability of Paternity above 99.99%. For an unrelated man, even a few non-matching loci drive the combined probability to essentially zero. The multiplication rule leaves almost no mathematical room for a middling result.
What does a Combined Paternity Index (CPI) of 100,000 mean in plain language?
A CPI of 100,000 means the DNA evidence is 100,000 times more consistent with the tested man being the biological father than with a random, unrelated man from the same population being the father. When converted to a Probability of Paternity using standard Bayesian calculation with neutral prior odds, a CPI of 100,000 yields a probability of approximately 99.999%.
Can a paternity test ever be inconclusive?
Yes, though it is uncommon. A result is considered inconclusive when the Combined Relationship Index falls between 0.1 and 10. This can happen in cases involving a mutation at one locus, a close biological relative as the possible alternative father, or degraded sample quality. When an inconclusive result occurs, USDC's laboratory can extend the analysis to test up to 28 genetic markers total, which resolves most cases definitively.
Does a 99.99% result mean there is a 0.01% chance the man is not the father?
That 0.01% figure reflects the mathematical structure of Bayesian probability reporting, not an actual 1-in-10,000 chance the result is wrong. The result assumes a neutral prior probability of 50% before testing. In practice, when the Combined Paternity Index is very large and all tested markers match, the scientific conclusion of paternity is extremely robust. The phrasing is a statistical convention, not an expression of meaningful doubt.
Is the science in a home paternity test the same as a legal paternity test?
Yes. The underlying lab science is the same as court-admissible testing. Both use STR analysis across the same type of genetic markers, the same population allele frequency databases, and the same probability calculations. The distinction is in sample collection: legal tests require witnessed collection and a documented chain of custody so results can be presented in court. Home kits involve self-collected swabs and are designed to give you answers for personal knowledge only.
How many markers does USDC's lab test?
USDC's laboratory analyzes up to 28 genetic markers. The standard analysis covers 23 autosomal STR markers. For cases that return an inconclusive Combined Relationship Index, the lab can run additional markers to bring the total up to 28, providing greater statistical confidence in the final result.
0 comments