Combined Paternity Index: What It Is and How to Read It

The Combined Paternity Index, or CPI, on a DNA test report is a single number showing how many times more likely the genetic results are if the tested man is the biological father than if a random, unrelated man of the same population were the father. A CPI of 1,000,000 means the DNA evidence is one million times more consistent with paternity than with an unrelated match. It is an odds ratio, not a percentage, and the higher it climbs above 1, the stronger the genetic support for paternity. Labs convert it into the Probability of Paternity percentage that also appears on the report, and under the Uniform Parentage Act a CPI of at least 100 paired with a 99 percent or higher probability creates a legal presumption of fatherhood.¹

How the Number Is Built

A paternity test examines specific locations on the chromosomes called loci. At each locus, the lab compares the child’s DNA to the alleged father’s and asks how much more likely the shared genetic pattern is if the tested man is the father versus if a random man is. The answer at that single marker is the individual Paternity Index (PI), calculated as a likelihood ratio: the probability of the child’s profile assuming the tested man is the father divided by the probability assuming an unrelated man is.²

The size of each PI depends on how common the shared allele is in the reference population. A shared allele that only 5 percent of people carry produces a much higher PI than one carried by 30 percent, because fewer random men could have passed on the rare variant.

The CPI is what you get when the lab multiplies every individual PI together. Modern tests analyze at least 20 STR (short tandem repeat) loci, so the CPI reflects the cumulative weight of all of them. Each additional matching locus multiplies the result higher, which is why more markers produce more definitive numbers.

Turning the CPI Into a Percentage

Most reports also list a Probability of Paternity as a percentage because it is easier to read than a raw odds ratio. Labs calculate it using Bayes’ theorem, combining the CPI with a prior probability that represents how likely paternity was before any DNA testing. That prior is almost always set at 0.50, a neutral starting point that treats paternity and non-paternity as equally likely before the evidence comes in.³

With a 0.50 prior, the conversion produces predictable numbers:

CPI of 100 gives roughly 99 percent probability of paternity
CPI of 1,000 gives about 99.9 percent
CPI of 1,000,000 gives about 99.9999 percent

The probability creeps closer to 100 percent as the CPI rises, but it never quite reaches it.³ A CPI of 500,000 does not translate to a 500,000 percent chance of anything; it means the DNA is 500,000 times more consistent with paternity than with a random unrelated match.⁴

Reading Your Result

A report will show one of two outcomes. If the tested man is included as the biological father, the CPI will typically sit in the hundreds of thousands, millions, or higher, paired with a Probability of Paternity above 99 percent. Results from accredited labs routinely exceed 99.99 percent for true fathers.⁴

If the tested man is excluded, the report will show mismatches at multiple loci. Labs generally require inconsistencies at a minimum of three or four markers before declaring an exclusion, because a mismatch at just one or two locations can be explained by a natural mutation rather than the absence of a biological relationship.⁵ A CPI between 0 and 1 means the genetic evidence actually favors non-paternity.⁴

What Makes the CPI Higher or Lower

Two reports for the same true father can produce very different CPIs. A handful of factors drive that variation.

Number of Markers Tested

Every additional locus that matches multiplies the CPI upward. Older tests examined fewer loci and produced smaller, though still conclusive, numbers. The current standard of at least 20 STR loci typically yields CPIs in the millions or billions for true fathers.

How Rare the Shared Alleles Are

Sharing uncommon alleles produces a much higher CPI than sharing common ones. Two tests using the same number of markers can land in very different ranges depending on which specific variants happen to be shared.

Whether the Mother Was Tested

Including the mother’s DNA lets the lab identify with certainty which allele the child inherited from her, leaving the paternal allele isolated. This generally produces a higher and more precise CPI. A test done without the mother, sometimes called a motherless or duo test, still works but tends to produce a more conservative number.

Genetic Mutations

A true biological father can show a mismatch at one or two markers because of a natural mutation. Labs account for this by requiring at least three inconsistencies before calling an exclusion, and they may run additional markers to clarify a suspicious pattern.⁵

Which Population Database the Lab Used

The allele frequencies feeding the PI calculation come from a reference database tied to a specific ethnic or racial group. An allele that is rare in one population may be common in another, which directly moves the CPI up or down. Under the Uniform Parentage Act, anyone who objects to the database the lab chose can request recalculation using a different population group within 30 days of receiving the report.¹

The Legal Threshold

The Uniform Parentage Act, adopted in some form by a majority of states, sets two genetic requirements for identifying a man as a child’s father: a Probability of Paternity of at least 99 percent, calculated using a prior probability of 0.50, and a combined index of at least 100 to 1.¹ A CPI of 100 with a 0.50 prior already produces roughly 99 percent probability, so meeting one threshold generally means meeting the other. Modern inclusion results almost always clear both bars by a wide margin.

Thresholds differ elsewhere. In Europe the accepted CPI cutoff is generally 1,000, while in the United States it can be as low as 100.⁴ Meeting or exceeding the applicable threshold creates a rebuttable presumption of paternity, which the identified father can challenge only by submitting to additional genetic testing that meets the same accreditation standards.¹

1
Administration for Children and Families. Uniform Parentage Act
2
National Institute of Justice. Population Genetics and Statistics for Forensic Analysts – Paternity Calculations
3
National Institute of Justice. Population Genetics and Statistics for Forensic Analysts – Probability of Paternity
4
National Institute of Justice. Population Genetics and Statistics for Forensic Analysts – Parentage and Relatedness
5
PubMed. Probability of Exclusion in Paternity Testing: Time to Reassess