Rewritten on: December 19, 2025
Our experts explain in detail the true meaning of the Random Match Probability (RMP) and probability of paternity that indicate the accuracy of DNA testing, the international AABB and ISFG standards, and the quality management required for trustworthy testing results.
DNA testing is used not only to confirm personal identity but also in a wide range of situations, including legal consultations, international procedures, inheritance matters, and criminal investigations. Even so, many people wonder, "Can it really be trusted?" or "What does 99.99% actually mean?" This article draws on the international standards of the AABB (American Association of Blood Banks) and ISFG (International Society for Forensic Genetics) to give an expert yet accessible explanation of how to correctly read the "accuracy" of DNA testing and the "probability of paternity."
- ・What "Accuracy" Means in DNA Testing
- ・The Basis for High Accuracy: Random Match Probability (RMP)
- ・The Relationship Between the Number of STR Marker Loci and Accuracy
- ・The "Probability of Paternity" Used in Paternity Testing
- ・The Probability of Paternity Is Not "the Probability of Being Parent and Child"
- ・International Standards (AABB, ISFG) and the Accuracy They Require
- ・Why Do Errors in DNA Testing Happen?
- ・Key Points for Reliable Testing Results
- ・Summary
What "Accuracy" Means in DNA Testing
The accuracy of DNA testing is an evaluation metric that shows how low the "statistical probability that an unrelated person happens to have the same DNA profile" (the random match probability) is. The lower this probability, the higher the test is considered to be in accuracy.
Modern forensic testing analyzes 20 or more STR (Short Tandem Repeat) sites simultaneously. STRs are regions scattered throughout the human genome where a sequence of about 2 to 6 base pairs repeats, and because the number of repeats differs from person to person, they are used for individual identification. [ref:1] Assuming the match probability at each locus is independent, multiplying the probabilities across all loci together yields the overall random match probability, so the more loci analyzed, the more dramatically the discriminating power improves. [ref:2]
The Basis for High Accuracy: Random Match Probability (RMP)
RMP (Random Match Probability) is the probability that an unrelated person happens to share the same DNA profile by chance. In forensic testing, the RMP typically reaches around 10⁻¹⁴ to 10⁻¹⁷, with 10⁻¹⁵ corresponding to "one in a quadrillion." This denominator far exceeds the world's population (about 8 billion), so the probability of finding another person with the same profile is statistically almost zero. [ref:1]
That said, there is a limitation: identical twins essentially share the same DNA sequence, so standard STR analysis cannot distinguish between them. The specific value of the RMP varies depending on the number of STR markers used and the regional allele frequency database applied, with standard datasets such as CODIS, managed by the FBI, and those from NIST widely used. [ref:1] [ref:8]
The Relationship Between the Number of STR Marker Loci and Accuracy
A key factor that determines the accuracy of DNA testing is the number of STR marker loci. Thirteen loci (the original CODIS standard) used to be the norm, but today kits such as GlobalFiler™ and PowerPlex® Fusion, which use 20 or more loci, are the mainstream. [ref:2]
| Number of STR loci | Approximate RMP range | Level of discriminating power |
|---|---|---|
| 13 loci | 10⁻¹⁰ to 10⁻¹³ | High |
| 20+ loci | 10⁻¹⁴ to 10⁻¹⁷ | Extremely high |
These figures are approximate and can vary depending on the ethnic background of the test subjects and which allele frequency database is used. However, the principle that increasing the number of loci dramatically improves discriminating power holds consistently, and analysis of 20 or more loci has become the international standard. [ref:9]
The "Probability of Paternity" Used in Paternity Testing
In paternity testing, results are expressed as the "Probability of Paternity." The basis for this calculation is the Paternity Index (PI), a likelihood ratio that compares the following two hypotheses.
- The hypothesis that the alleged father is the true biological father (H₁)
- The hypothesis that a randomly selected, unrelated man is the father (H₂)
The final probability of paternity is obtained by multiplying the PI values from each STR locus to get the CPI (Combined Paternity Index), then combining it with a prior probability (typically 0.5) and applying Bayes' theorem. [ref:3] [ref:4]
W = CPI / (CPI + 1)
For example, if the CPI is 10,000, then W ≈ 99.99%. The larger the CPI, the closer the result approaches 100%, but in principle it never reaches exactly 100%.
The Probability of Paternity Is Not "the Probability of Being Parent and Child"
- A probability of paternity of 99.99% does NOT mean a 99.99% probability of an actual biological parent-child relationship.
- Nor does it mean there is a 0.01% chance the two are not parent and child.
- The figure can change depending on how the prior probability is set.
This figure is a statistical measure of "how strongly the DNA data supports the hypothesis," and the ISFG recommends that testing reports present the likelihood ratio as the primary figure. [ref:3] When the probability of paternity reaches 99.9% to 99.99%, it is judged to be virtually impossible to deny the biological parent-child relationship. When a parent-child relationship is denied, allele mismatches are confirmed at multiple loci, and the result is clearly classified as an "exclusion." [ref:5]
International Standards (AABB, ISFG) and the Accuracy They Require
The AABB (Association for the Advancement of Blood & Biotherapies) is one of the world's most trusted accrediting bodies for paternity testing laboratories. Accredited labs undergo rigorous review of the following items. [ref:4]
- Standardization and validation of DNA analysis procedures
- Chain of custody (management of sample collection, transport, and storage)
- Ensuring the validity and reproducibility of statistical analysis
- Regular participation in proficiency testing (external quality assurance programs)
Under most international standards, legal paternity testing requires a probability of paternity of at least 99%, and 99.9% or higher at the practical level. The U.S. Citizenship and Immigration Services (USCIS) also requires at least 99.5%, a threshold set as an administrative acceptance requirement. The ISFG regularly revises and publishes statistical analysis guidelines, promoting the standardization of statistical processing using likelihood ratios. [ref:3]
Why Do Errors in DNA Testing Happen?
Even though the scientific accuracy of DNA testing is extremely high, "errors" are sometimes reported in the media. In most cases, the cause is not the DNA analysis technology itself but human factors or flaws in the process, such as the following. [ref:5] [ref:10]
- Sample mix-ups: Cases where labels are mistakenly swapped during sample collection or transport
- Contamination: Cases where another person's DNA is mixed into a sample
- Chain-of-custody failures: Cases where a break in the management record makes it impossible to rule out tampering
- Statistical analysis errors: Cases involving use of an inappropriate allele frequency database or incorrect assumptions in the calculation model
- Misinterpretation of results: Cases where the statistical meaning is not correctly understood, leading to a wrong conclusion
These problems are greatly reduced in environments where external audits and proficiency testing are mandatory, such as at AABB-accredited laboratories.
Key Points for Reliable Testing Results
No matter how high the statistical accuracy, the reliability of the result cannot be maintained if the following elements are missing.
- Proper sample collection procedures: Samples must be collected correctly with identity verification and a witness present; legal testing requires ID verification and photographs.
- Maintaining chain of custody: The management record must remain unbroken from the time the sample is collected until the results are produced.
- Lab quality management and international accreditation: At labs accredited under international standards such as ISO 17025 or AABB, the entire operational process is regularly audited by a third party.
Procedural reliability matters just as much as scientific accuracy. When choosing a testing organization, be sure to check its quality management system and whether it holds international accreditation, not just its marketing claims about accuracy.
The seeDNA Genetic Medical Research Institute has obtained the international quality standard ISO 9001 certification and the Privacy Mark for privacy protection. We adhere to strict quality management procedures across all our DNA testing, ensuring you can receive your results with complete peace of mind.
Summary
- DNA testing is a highly accurate examination that, through analysis of multiple STR sites, achieves an extremely low random match probability. With analysis of 20 or more loci, the RMP reaches 10⁻¹⁴ or lower.
- The probability of paternity is not "the probability of being parent and child," but rather a statistical measure of how strongly a hypothesis is supported.
- Under international standards (AABB, ISFG), 99.9% or higher is generally required as the baseline.
- "Errors" in DNA testing stem more from human factors and process flaws than from the analysis technology itself.
- Lab accreditation, procedures, and quality management are essential to the reliability of results.
Apply / Purchase a KitFrequently Asked Questions
Q1. Specifically how accurate is DNA testing?
A. Modern forensic DNA testing, which analyzes 20 or more STR loci, achieves a Random Match Probability (RMP) of about 10⁻¹⁴ to 10⁻¹⁷. This is far lower than one in a trillion, making it possible to identify individuals with extremely high accuracy.
Q2. Does a "99.99% probability of paternity" mean "a 99.99% probability of being parent and child"?
A. No, strictly speaking these are different things. The probability of paternity is a statistical measure of "how much more strongly the hypothesis that the alleged father is the biological father is supported compared to the hypothesis that another man is the father." It is a calculated value based on a prior probability of 0.5, and it does not directly represent an "absolute probability of being parent and child." That said, when this value is 99.9% or higher, the parent-child relationship is considered virtually certain in scientific and practical terms.
Q3. Can errors occur in DNA testing?
A. Because the accuracy of the DNA analysis technology itself is extremely high, technical errors are highly unlikely. However, the possibility of human-factor errors — such as sample mix-ups, contamination (mixing in of another person's DNA), and chain-of-custody failures — is not zero. These risks can be greatly reduced by choosing an organization with a rigorous quality management system, such as an AABB-accredited laboratory.
Q4. What are AABB and ISFG?
A. AABB stands for "Association for the Advancement of Blood & Biotherapies," and it is globally recognized as an international accrediting body for paternity testing laboratories. ISFG stands for "International Society for Forensic Genetics," an international academic organization that promotes research in forensic genetics and the standardization of statistical analysis methods.
Q5. Can DNA testing distinguish between identical twins?
A. Standard STR analysis cannot distinguish between identical twins, since they share the same DNA sequence. In recent years, identification methods based on whole-genome sequencing and detection of somatic mutations have been studied, but these are not currently part of standard paternity testing.
Q6. What should I check when choosing a DNA testing organization?
A. We recommend checking: (1) whether it holds international accreditation such as AABB or ISO 17025; (2) whether a chain of custody is established; (3) whether the report clearly states the statistical basis (paternity index or probability of paternity); and (4) whether it has personal data protection measures in place (such as a Privacy Mark). The seeDNA Genetic Medical Research Institute holds ISO 9001 and Privacy Mark certification, so you can use our services with confidence.
seeDNA Genetic Medical Research Institute's Peace-of-Mind Support
The seeDNA Genetic Medical Research Institute is a trusted, reliable DNA testing and genetic testing specialist that has obtained the international quality standard ISO 9001 and the Privacy Mark for privacy protection.
If you have concerns about family or parent-child biological relationships, a partner's infidelity, or similar issues, our DNA testing experts are here to support you with peace of mind — please feel free to contact us.
[Free Consultation with Specialist Staff]
If you have any questions,
please feel free to contact our toll-free number.
／Open every day, including weekends／
Business hours: Monday–Sunday 9:00 AM–6:00 PM
(excluding public holidays)
Author
Ph.D. / Test Analyst: L. L.
After earning a doctoral degree from the Graduate School of Clinical Medicine at the International University of Health and Welfare, joined seeDNA as a test analyst.
Responsible for prenatal parent-child DNA testing and data analysis.