There is no single percentage that makes a DNA test “positive,” because DNA tests are used for radically different purposes, and each purpose has its own threshold. A paternity test uses a probability calculation where anything above roughly 99% is treated as confirmation. A prenatal screening test needs the fetal DNA in a mother’s blood to reach at least about 5% of the total sample to be reliable. A cancer liquid biopsy might flag tumor DNA that makes up less than 1% of circulating DNA. The word “positive” itself means something different in each context, and so does the number attached to it.
Paternity Testing and the 99% Standard
Paternity tests are probably the most common reason people ask about DNA percentages, and they use a metric called the “probability of paternity” or “likelihood of paternity.” Most accredited labs and court systems treat a probability above 99% as a confirmation that the tested man is the biological father. Some jurisdictions set the bar at 99.0%, others at 99.9%, but the general threshold hovers in that range.
What many people do not realize is that this number is not a simple percentage of shared DNA. It is a statistical calculation comparing how likely the observed genetic overlap would be if the man were the father versus if a random unrelated man from the same population were tested. The output looks like a percentage, but it is really a probability statement about two competing explanations for the same data.
Older paternity tests, which relied on blood-type markers and a limited set of genetic markers rather than modern DNA profiling, were less decisive. A study examining these older methods found that while about 97% of men who were not the father could be excluded outright, the remaining few who could not be excluded sometimes had calculated paternity probabilities ranging anywhere from about 1% to nearly 99%.
1PubMed. Limitations of paternity testing calculationsThat ambiguity is largely a thing of the past. Modern DNA paternity tests examine many more genetic markers and routinely produce probabilities above 99.99% for true fathers, making false inclusions vanishingly rare. Noninvasive prenatal paternity tests, which work from a blood draw during pregnancy rather than waiting until birth, use different statistical machinery but aim for the same kind of decisiveness. One method sets a stringent threshold: paternity is confirmed only when the statistical evidence against a random match clears a very high bar, and cases that fall in a gray zone are reported as indeterminate rather than forced into a yes-or-no answer.
2Genetics in Medicine. Informatics-based, highly accurate, noninvasive prenatal paternity testingPrenatal Screening and the Fetal Fraction
Noninvasive prenatal testing, commonly called NIPT, screens for chromosomal conditions like Down syndrome by analyzing fragments of fetal DNA circulating in the pregnant person’s blood. The critical number here is the “fetal fraction,” meaning what percentage of the total cell-free DNA in the blood sample actually comes from the placenta and represents the fetus. If that fraction is too low, the test cannot reliably distinguish a chromosomal abnormality from background noise.
Research has established that a fetal fraction of about 5% is the minimum needed for dependable detection of trisomy 21, the most commonly screened condition. Below that level, the chance of a false negative climbs sharply.
3PubMed Central. Analysis of fetal fraction in non-invasive prenatal testing with low-depth whole genome sequencing For smaller chromosomal deletions, the situation is trickier. One study modeling microdeletion detection found that at a 10% fetal fraction, accuracy for detecting deletions reached about 79%, and it improved to over 98% only when the search was limited to deletions larger than 3 million base pairs.
4PLOS ONE. Non-invasive prenatal testing (NIPT) by low coverage genomic sequencing: Detection limits of screened chromosomal microdeletionsFetal fraction varies by gestational age, maternal weight, and other factors. Heavier individuals tend to have lower fetal fractions because they have more of their own cell-free DNA diluting the fetal signal. When a lab returns a “no result” or “test failure” rather than a positive or negative, it usually means the fetal fraction was too low to call. This is not the same as a negative result, and the person is typically advised to retest later in pregnancy when the fetal fraction has risen.
Cancer Detection Through Liquid Biopsy
Liquid biopsies look for fragments of tumor DNA, called circulating tumor DNA or ctDNA, floating in a patient’s bloodstream. The percentage of cell-free DNA that comes from the tumor varies enormously. In one study of cancer patients, ctDNA ranged from as little as 0.006% to over 90% of total circulating DNA.
5PubMed. Identification of circulating tumour DNA (ctDNA) from the liquid biopsy results: Findings from an observational cohort study A separate large study of advanced prostate cancer found a median ctDNA fraction of 5%, with the range spanning from undetectable to about 89%. Researchers in that study considered anything below 2% as effectively undetected by their methods.
6Nature Communications. Prediction of plasma ctDNA fraction and prognostic implications of liquid biopsy in advanced prostate cancerThe challenge is that ctDNA is highly fragmented and often present at extremely low concentrations against a large background of normal DNA, making detection of clinically relevant mutations difficult.
7PubMed Central. Techniques of using circulating tumor DNA as a liquid biopsy component in cancer management There is no universal percentage threshold for a “positive” liquid biopsy the way there is for a paternity test. Instead, the threshold depends on the specific assay’s sensitivity, the mutation being searched for, and how the results will be used clinically. Early-stage cancers shed far less DNA into the blood than advanced cancers, so the detection floor matters enormously for screening purposes. A test sensitive enough to catch 0.01% ctDNA fractions might find early tumors that a less sensitive assay would miss entirely.
Genealogy Tests and Shared DNA Between Relatives
Consumer DNA tests from ancestry companies report how much DNA you share with a match, usually expressed as a percentage or in centimorgans. The question “what percentage means we’re related?” comes up constantly, and the honest answer is that it depends on how distantly related you might be.
Close relatives share large, obvious amounts of DNA. A parent and child share roughly 50%. Full siblings share around 50% as well, though with more variation. First cousins share around 12.5% on average. These close relationships are easy for any test to detect with high confidence. The percentages get less predictable as you move further out. Third cousins share only about 0.78% on average, and the variation around that average is wide. Research on large datasets shows that about 98.5% of third-cousin pairs still share at least one detectable segment of DNA, so the test will usually spot them.
8PubMed Central. The rate of identical-by-descent segment sharing between close and distant relativesBut as you reach fifth cousins, only about a third of pairs share any detectable DNA at all. By eighth cousins, the probability drops to less than 1%.
8PubMed Central. The rate of identical-by-descent segment sharing between close and distant relatives This means that if a genealogy test shows zero shared DNA with someone, it does not prove you are unrelated. You could easily be fifth or sixth cousins whose shared segments simply were not passed down through the generations. DNA inheritance is random at each step, and distant connections frequently vanish from the detectable record.
Another complication is that the amount of DNA shared does not map neatly onto a single relationship type. Two people sharing 7% of their DNA could be half-first cousins, first cousins once removed, or even a great-uncle and great-niece. Ancestry companies provide a range of possible relationships, not a definitive answer, precisely because the same percentage is consistent with multiple family tree configurations.
When Background Sharing Muddies the Signal
Population background adds another layer of noise. In some populations, especially smaller or historically isolated ones, unrelated individuals share more DNA than you would expect by chance, simply because they descend from a limited pool of ancestors. Research on ethnolinguistically defined populations found that background DNA sharing increases enough in these groups that, beyond about fifth cousins, the shared DNA from a genuine genealogical relationship becomes impossible to distinguish from the elevated baseline.
9PLoS ONE. Cryptic Distant Relatives Are Common in Both Isolated and Cosmopolitan Genetic SamplesEven in large cosmopolitan populations, “cryptic” relatives are surprisingly common. Two strangers whose families have lived in the same broad region for centuries may share small DNA segments that look like a distant cousin match but actually reflect very deep shared ancestry. Sorting genuine recent relatives from this background noise requires looking not just at total shared DNA but at the length and pattern of shared segments. A method that estimates the proportion of the genome shared identically by descent within sliding windows across the genome can help distinguish true relatives from people who just happen to share population-level similarity.
10PLoS Genetics. Inference of Relationships in Population Data Using Identity-by-Descent and Identity-by-StateForensic genealogy tools have been developed to handle relationship inference across a wide range of kinship degrees, performing well against benchmarks for relationships out to the ninth degree when tested across diverse population backgrounds.
11BioTechniques. Analytical validation of the IBD segment-based tool KinSNP® for human identification applicationsMosaicism and the Problem of Low-Level Variants
Some DNA tests look for mutations that are not present in every cell of the body. Genetic mosaicism occurs when a mutation arises after fertilization, so only some of the body’s cells carry it. Detecting these mosaic variants requires a test sensitive enough to pick up a signal that might represent only a small fraction of the cells in a sample.
Traditional Sanger sequencing, the workhorse of clinical genetics for decades, struggles with mosaic variants below about 15 to 50% of the sample depending on the specific mutation. More modern next-generation sequencing can push the detection floor down to around 1%, but only with sufficient depth of coverage, meaning the same stretch of DNA is read many times over to build up a reliable signal.
12PubMed. Parallel sequencing used in detection of mosaic mutations: comparison with four diagnostic DNA screening techniquesThis matters clinically. A person with a mosaic mutation at 5% of their cells might show no symptoms, while the same mutation at 30% might cause disease. Whether the test “sees” the variant at all depends on the technology being used and the coverage depth. A lab reporting a negative result on a standard sequencing test cannot rule out low-level mosaicism. Specialized deep-sequencing panels are needed when mosaicism is suspected.
Mitochondrial DNA and the Heteroplasmy Threshold
Mitochondrial DNA operates by different rules than the nuclear DNA most people think of when they hear “DNA test.” Each cell contains hundreds or thousands of copies of mitochondrial DNA, and not all of those copies are necessarily identical. When a person carries a mix of normal and mutant mitochondrial DNA, that state is called heteroplasmy, and the crucial question is what percentage of the mitochondrial copies carry the mutation.
Pathogenic mitochondrial mutations often only cause disease when they exceed a certain proportion of the total. Below that threshold, the normal copies compensate. A systematic review examining this biochemical threshold found evidence that cells with mutant mitochondrial DNA fractions below about 60% could still show reduced activity of the cellular energy machinery, suggesting the critical level may be lower than the 70 to 90% range traditionally cited in textbooks.
13PubMed Central. A systematic review on the biochemical threshold of mitochondrial genetic variants The threshold varies by mutation type, tissue, and individual. Some mutations cause problems at 50% heteroplasmy; others are tolerated at 80%. Wild-type and mutant copies coexist, and disease manifests only when the pathogenic fraction pushes past whatever threshold applies.
14PubMed Central. Mitochondrial DNA heteroplasmy in disease and targeted nuclease-based therapeutic approachesFor diagnostic purposes, a lab might report a heteroplasmy level of, say, 45%, but interpreting whether that is “positive” in a clinical sense requires context about which mutation it is and which tissue was sampled. Muscle biopsies often reveal higher heteroplasmy levels than blood samples from the same person because the mutation may accumulate differently in different tissues.
Forensic DNA Profiling
In criminal forensics, the question is not “what percentage of DNA is shared?” but “how strongly does this DNA profile match the suspect?” The result is usually expressed as a likelihood ratio: how many times more likely is it to observe this DNA evidence if it came from the suspect versus a random person. Likelihood ratios in strong forensic matches can be astronomically large, in the billions or trillions, because modern profiling systems examine enough genetic markers to make coincidental matches essentially impossible.
15PubMed. Likelihood ratios for DNA identificationThere is no fixed percentage threshold for declaring a forensic match. Instead, the strength of the evidence is expressed as that ratio, and courts or investigators evaluate it alongside other evidence. When the DNA sample is degraded or mixed with DNA from multiple people, the analysis becomes much harder. The likelihood ratio might drop from the trillions to the thousands or even lower, and interpretation shifts from a near-certainty to a probabilistic statement that requires careful communication to juries.
Chimerism and When One Body Has Two DNA Profiles
A small number of people are chimeras, meaning they carry cells with two distinct DNA profiles. This can happen naturally through events during early embryonic development, or artificially after a bone marrow or stem cell transplant. After a transplant, the recipient’s blood cells gradually become the donor’s genetic type, so a DNA test on blood might return a completely different profile than a test on a cheek swab or skin sample from the same person.
16PubMed. Forensic implications of the presence of chimerism after hematopoietic stem cell transplantationThis creates real problems for paternity testing, forensic identification, and even ancestry results. A chimeric individual might fail a paternity test for their own biological child if the wrong tissue is sampled. In forensic contexts, blood left at a crime scene by a transplant recipient could point investigators toward the bone marrow donor rather than the actual person who was there. Labs that are aware of a transplant history can work around the issue by testing a tissue type that was not affected by the transplant, but the chimeric state has to be known about or suspected in the first place.
Gene Editing Verification
An emerging application of DNA percentage thresholds involves verifying how well a gene-editing tool like CRISPR has worked. After editing cells, researchers need to know what fraction of the target DNA was successfully modified. One quantitative method showed it could measure editing efficiencies in the range of 43 to 49% at the intended target sites, while also confirming that unintended “off-target” editing at other genomic locations was not detectable above the noise floor. The method’s practical sensitivity limit was about 5%, meaning edits occurring in fewer than 5% of copies could potentially be missed.
17Nucleic Acids Research. qEva-CRISPR: a method for quantitative evaluation of CRISPR/Cas-mediated genome editing in target and off-target sitesThis matters for therapeutic gene editing, where safety depends on showing that the editing tool modified the right gene without also introducing unwanted changes elsewhere. In that context, “positive” for off-target editing could mean detecting even a tiny percentage of unintended modifications, and the sensitivity of the verification method sets the floor for what can be ruled out. A method that cannot see changes below 5% cannot guarantee that low-level off-target edits did not occur.
Ancient DNA and Contamination Thresholds
When researchers sequence DNA from ancient bones, a major concern is modern human contamination. A bone fragment thousands of years old might yield DNA that is, say, 15% ancient and 85% modern contamination from handling. Determining whether the ancient fraction is large enough to trust requires specialized approaches. Researchers working on the Neanderthal genome argued that the only reliable way to estimate modern contamination is to look directly at positions in the DNA where Neanderthals are known to differ from all living humans. Other indirect measures, like checking how fragmented the DNA is or looking at characteristic chemical damage patterns, turned out to be unreliable on their own.
18PubMed Central. The Neandertal genome and ancient DNA authenticitySimilarly, when analyzing ancient genomes for segments shared with other ancient individuals, false positive signals can arise in regions of the genome with sparse data coverage. Filtering out genomic regions with low marker density and masking areas around centromeres and telomeres, which together cover roughly 8% of the genome, helps reduce these spurious signals.
19Nature Genetics. Accurate detection of identity-by-descent segments in human ancient DNA The threshold for calling an ancient DNA result “positive” is therefore less about a single percentage and more about whether enough quality-controlled data exists to make a confident call after known sources of error have been stripped away.
Infectious Disease PCR Tests
PCR-based tests for infections, like those used during the COVID-19 pandemic, amplify tiny amounts of pathogen DNA or RNA until it crosses a detectable threshold. The key measurement is the cycle threshold, or Ct value: the number of amplification cycles needed before the target genetic material is detectable. A lower Ct value means more pathogen was present in the original sample; a higher Ct value means less was there.
Labs set a Ct cutoff, often around 35 to 40 cycles depending on the assay, above which the result is considered negative. Setting that cutoff involves balancing sensitivity against specificity. A very generous cutoff detects extremely low levels of pathogen but risks picking up contamination or non-viable genetic fragments. A stricter cutoff misses some true low-level infections. The choice of cutoff depends on whether the goal is diagnosis, screening, or epidemiological surveillance, and variability in amplification efficiency between test runs can shift the effective cutoff slightly, requiring standardization procedures.
20PubMed. Selection of a cutoff value for real-time polymerase chain reaction results to fit a diagnostic purpose: analytical and epidemiologic approachesUnlike the other DNA tests discussed here, a PCR result is typically reported as simply positive or negative, with the underlying Ct value available to the ordering clinician but rarely shared with the patient. The percentage concept does not apply in the same way; the question is whether the pathogen’s genetic material was present above a defined detection floor, not what fraction of the total sample it constituted.