When a sibling DNA test result arrives, most people look at the words "supports" or "does not support" a full-sibling relationship and stop reading. But the number that actually drives that verdict is called the Combined Sibling Index, or CSI. Unlike a paternity test, where a match usually produces a Probability of Paternity above 99.99 percent and the conclusion feels definitive, a sibling test lives in a wider gray zone. Understanding the CSI is the difference between reading your report and actually interpreting it.
This guide walks through what the CSI number means, how the lab calculates it, why sibling numbers are almost always lower than paternity numbers, and what CSI ranges typically map to which conclusions.
What the Combined Sibling Index Actually Is
The Combined Sibling Index is a likelihood ratio. Like the Combined Paternity Index used in paternity tests, it compares two possible explanations for the DNA patterns the lab observes:
- The two tested individuals are full biological siblings (share both biological parents).
- The two tested individuals are unrelated people from the same general population.
A CSI of 50 means the DNA evidence is 50 times more likely to be seen if the two people are full siblings than if they are unrelated strangers. The number is a ratio of two probabilities, not a percentage. Reports sometimes label this same value as a "Combined Sibship Index" or a "Full Sibling Index" — the math is identical.
Half-sibling tests use the same framework but compare "half siblings share one parent" against "unrelated." The resulting Combined Half-Sibling Index is interpreted on the same scale as the CSI, though the underlying probabilities differ because half siblings on average share only about 25 percent of their DNA versus 50 percent for full siblings.
Why Sibling Numbers Are Smaller Than Paternity Numbers
A paternity test compares a straightforward biological question: did this specific man contribute one of the child's two alleles at each locus? Because a father must have passed one allele at every location, the math is highly constrained and the Combined Paternity Index can quickly reach the millions or higher when every marker matches.
A sibling test asks a fuzzier question: are the DNA patterns of these two people more consistent with them sharing two parents, one parent, or none? Full siblings share about half their DNA on average, but the amount actually shared at any given locus varies from zero to two alleles. Two full siblings might share both alleles at one marker, one allele at another, and neither at a third — all consistent with the sibling relationship. This natural variation means the statistical evidence per marker is weaker than in a paternity test, and the CSI ceiling is much lower.
In practice, CSI values above 100 provide strong support for full siblings. Values above 1,000 are considered very strong. By comparison, a Combined Paternity Index of 100 is barely meeting the legal minimum in most states, and a paternity report often shows a value in the millions. Do not compare a CSI to a CPI expecting the same scale.
How the Lab Builds the CSI, Marker by Marker
Modern relationship testing uses Short Tandem Repeat, or STR, analysis. The lab examines specific regions of DNA where a short pattern of letters repeats a variable number of times. At each region, called a locus, a person carries two alleles, one inherited from each biological parent. The NIST Short Tandem Repeat DNA Internet Database catalogs the STR markers used in forensic and relationship testing worldwide and publishes the allele frequency data labs rely on.
For every locus tested, the lab compares the two tested individuals and calculates a Sibling Index for that marker. The Sibling Index at one locus is a mini likelihood ratio that answers a narrow question: at this single marker, how much more likely is the observed allele-sharing pattern if these two people are full siblings than if they are unrelated?
The Combined Sibling Index is then the product of all the individual Sibling Indices multiplied together. If a test examines 23 markers and produces per-locus indices ranging from 0.4 to 3.8, the CSI is the product of all 23 values. Because some markers can produce a per-locus index below 1 (evidence weakly against siblings), the multiplication can push the total up or down. This is another reason CSI numbers are smaller and more variable than CPI numbers, where the per-locus indices are almost always above 1 when paternity holds.
Why Rare Alleles Change Everything
Population allele frequencies are the hidden lever in every Sibling Index calculation. Labs use published reference databases, often organized by ethnic background, to estimate how common each allele is in the broader population. Rare alleles are worth much more evidentially than common ones.
Consider two people who both carry an allele at a particular locus. If that allele occurs in 40 percent of the population, the sharing is only mildly informative — it could easily happen by chance in two unrelated people. If the shared allele occurs in only 2 percent of the population, the same match becomes much stronger evidence that the two people inherited it from a common parent rather than by coincidence.
This is why two sibling tests that both reach "supports full siblings" can produce very different CSI numbers. Two children who each share the expected pattern with their siblings might land at CSI 15 versus CSI 5,000 depending entirely on how common or rare their inherited alleles happen to be in the reference population.
Converting CSI to a Probability of Siblingship
Some sibling reports also express the result as a Probability of Siblingship, a percentage figure calculated from the CSI using the same Bayesian framework used in paternity reports:
Probability of Siblingship = CSI / (CSI + 1) x 100
The formula uses a prior probability, which represents what you would assume before seeing any DNA evidence. Standard practice, endorsed by the International Society for Forensic Genetics, is to use a neutral 50/50 prior — meaning the calculation starts by assuming the two people are equally likely to be full siblings or unrelated, and then the DNA evidence updates that assumption.
Under a 50/50 prior, a CSI of 10 gives a Probability of Siblingship of about 91 percent. A CSI of 100 gives about 99 percent. A CSI of 1,000 gives 99.9 percent. The higher the CSI, the closer the probability approaches, but never reaches, 100 percent. Unlike paternity, where the same math produces 99.99 percent or higher in almost every confirmed case, sibling probabilities often land in the 90-to-99.9 percent range even when the biological relationship is real.
CSI Values You Are Likely to See on Your Report
Here is a reference table showing common CSI values, the corresponding Probability of Siblingship under a neutral 50/50 prior, and the typical interpretation.
| CSI Value | Probability of Siblingship | Typical Interpretation |
|---|---|---|
| Below 1 | Below 50% | Evidence does not support full siblings. The observed DNA patterns are more consistent with unrelated individuals. |
| 1 to 10 | 50% to 91% | Inconclusive. The evidence leans toward siblings but is not strong enough for most labs to report a definitive verdict. |
| 10 to 100 | 91% to 99% | Moderate support for full siblings. Most labs will report this as "supports a full sibling relationship" with a stated confidence level. |
| 100 to 1,000 | 99% to 99.9% | Strong support. This is the typical range for a clear full-sibling result on a standard STR panel. |
| Over 1,000 | Over 99.9% | Very strong support. Often seen when a known parent's DNA is included or when the shared alleles happen to be rare. |
Notice how much more the outcome depends on the specific CSI value than in a paternity test. A paternity result of "confirmed" almost always sits above 99.99 percent. A sibling result of "supports full siblings" might land anywhere from 91 percent to well above 99.9 percent, and where it lands changes how confident you should be in the conclusion.
How to Strengthen a Sibling Test Result
Sibling tests are inherently statistical, and a borderline CSI can be pushed one direction or the other by adding more information. Three practical options can raise a marginal result into a clear one:
- Add a known parent's sample. If a mother, father, or other shared parent is available, including them dramatically improves the calculation. The lab can then determine which alleles came from which parent, turning ambiguity into direct genetic accounting. This is the single biggest improvement most sibling tests can make.
- Test on a larger marker panel. The more STR loci analyzed, the more independent Sibling Indices contribute to the total. USDC's home sibling test uses up to 28 genetic markers per person. The standard panel is 23 autosomal STR markers — above the industry baseline of 20 or more markers — and the lab can add up to 5 more unique autosomal markers to reach 28 total when a kinship result comes back inconclusive. The additional markers are not run on every case; they are added when the standard panel does not resolve the relationship clearly. When they are added, the resulting CSI is larger and more stable.
- Test additional siblings. If more than one alleged sibling is available, testing three or more people together provides cross-checks the lab can use to resolve ambiguity that would otherwise stall a two-person test.
None of these are always possible — the whole reason many people order a sibling test is that the parent is unavailable, unknown, or deceased — but when any of them are practical, they meaningfully change what the report can conclude.
Full Sibling vs Half Sibling: Which Test to Order
If you already suspect the answer might be "half siblings share one parent" rather than "full siblings share both parents," ordering the correct test up front matters. The lab compares the observed DNA sharing against the specific hypothesis you asked about. A full-sibling test that produces a low CSI might not mean the two people are unrelated — it might mean they are half siblings, and the full-sibling hypothesis is being tested against the wrong alternative.
USDC offers a home full sibling DNA test and a home half sibling DNA test, both at $139. If the biological relationship in question is uncertain — you know these two people share at least one parent but you are not sure whether it is one or both — the full-sibling test is usually the right starting point, because a strong full-sibling CSI rules in both parents, while a weak full-sibling CSI at least tells you the "both parents shared" hypothesis is not well-supported and half-sibling testing may be warranted next. A grandparent DNA test is another route in cases where the alleged shared parent is unavailable but that parent's own parents are.
What About Legal Sibling Testing?
The laboratory science is identical for home sibling tests and legal sibling tests. The same STR markers, the same per-locus Sibling Indices, the same multiplication into a CSI, and the same population reference databases apply to both. The difference is chain of custody.
A legal sibling test requires witnessed sample collection with verified identities so the result is admissible in a court proceeding, an inheritance case, or an immigration filing. A home test skips that step because the result is for personal knowledge only. If you know a court, agency, or attorney will need to rely on the result, a chain-of-custody test is the correct route from the start — retrofitting a home result into a legal record after the fact is not possible.
Educational context on the AABB's accreditation program for relationship testing facilities is available on the AABB accreditation overview, which describes the standards labs must meet to perform court-admissible relationship testing.
What to Look for on Your Own Sibling Report
When your sibling report arrives, here is a practical walkthrough of what to check.
- The overall conclusion. The report will state whether the DNA supports a full sibling relationship, a half sibling relationship, or does not support the tested relationship. It may also note that the result is inconclusive and recommend further testing.
- The Combined Sibling Index. This is the large ratio, often ranging from below 1 (weak or against) to the thousands (strong support). A larger CSI means stronger statistical support. Check where your value falls in the reference table above to gauge how firm the conclusion actually is.
- The Probability of Siblingship. Not every report includes this percentage — some report only the CSI. If it is present, it converts the CSI to a probability using a 50/50 prior. A probability of 99 percent or higher usually means the lab is comfortable calling the relationship; below 90 percent usually reads as "leans one way, but cannot conclude."
- The individual marker table. Most reports include the alleles observed at each locus for both tested individuals. This is where you can spot which markers contributed most to the CSI and whether any single locus produced an unusual result. A definition of allele and locus is available in the NIH Talking Glossary of Genomic and Genetic Terms.
- The prior probability used. Reports usually state this explicitly, most often 0.5, or 50/50. If a different prior is used it should be labeled clearly.
- Whether a parent was tested. If a shared parent was included, the report will say so, and the CSI will generally be higher than a two-person test. If not, and the result is inconclusive, adding a parent is the most common next step.
USDC's home sibling test, priced at $139, produces a full statistical report with the CSI, per-locus Sibling Indices, the prior used, and a plain-English conclusion. Lab processing takes 2 to 3 business days from sample receipt, and the total order-to-results window is 7 to 10 days with standard shipping. Kits and results are also available through the full home DNA test collection. If you are still deciding which relationship test to order, the how DNA relationship testing works page compares the options.
Frequently Asked Questions
What is a good CSI number on a sibling DNA test?
A CSI above 100 usually corresponds to a Probability of Siblingship above 99 percent and is considered strong support for the tested relationship. Values between 10 and 100 are moderate support. Values below 10 are generally treated as inconclusive, and values below 1 lean against the tested relationship. Unlike a paternity test where the number often reaches the millions, sibling tests rarely produce CSI values that high because the underlying biology is less deterministic.
Why is my sibling test result inconclusive if I know we have the same father?
Full siblings share about half their DNA on average, but the actual amount at any given marker varies. Two true full siblings can, purely by chance, share fewer alleles than a two-person test can definitively call. The single most effective fix is including a known parent's sample, which lets the lab account for which alleles came from where. Adding a third sibling or upgrading to a lab that tests more markers can also help.
Can a sibling test tell the difference between full and half siblings?
Yes, but not always cleanly with only two people. A dedicated full-sibling test and a dedicated half-sibling test compare the DNA against different hypotheses, and the results together can distinguish between the two scenarios. When a two-person test comes back ambiguous, some labs will run both calculations and report which relationship is most consistent with the data. Adding a shared parent, or testing more siblings, produces the cleanest resolution.
Do court-admissible sibling tests give a higher CSI than home tests?
No. The laboratory analysis is identical for home tests and court-admissible legal tests. The same STR markers, per-locus Sibling Indices, multiplication into a CSI, and reference databases are used. The only difference between the two types is chain of custody — who witnesses the collection and how identities are verified. Chain of custody makes the report admissible in court but does not change the underlying math.
What if my CSI is below 1?
A CSI below 1 means the observed DNA sharing is more consistent with the two people being unrelated than with the tested relationship. This is not the same as a definitive exclusion — it simply means the evidence weighs against the hypothesis you tested. If a shared parent could not be included in the original test, adding one is the most common next step before drawing a firm conclusion. The how DNA relationship testing works page covers what to expect from the process end to end.
0 comments