First cousins share roughly 12.5 percent of their DNA on average, but the actual amount in any given pair can swing several percentage points above or below that figure. The reason is that DNA is not passed down in neat, fixed portions. Recombination shuffles the deck at every generation, so two first cousins might share 10 percent of their genome or 15 percent, even though the textbook prediction is one-eighth. That randomness only grows as you move to more distant relationships, and by the time you reach fifth or sixth cousins, many pairs share no detectable DNA at all.
What the Theoretical Numbers Look Like
The expected amount of shared DNA follows a straightforward halving pattern. You get half your DNA from each parent, so siblings share about 50 percent. First cousins, who share one set of grandparents, are expected to share about 12.5 percent. Second cousins (sharing great-grandparents) drop to around 3.125 percent. Third cousins land near 0.78 percent. Each additional step of cousinhood roughly cuts the expected sharing in half again.
These numbers describe averages across many pairs. They tell you the probability that any single spot in the genome was inherited from the same ancestor. At any one location in the genome, two first cousins either share a copy from their common grandparent or they don’t. Across the whole genome, the law of large numbers pulls the total toward 12.5 percent, but it does not land there exactly. The standard deviation for first cousins is around 4 percentage points, meaning a pair of first cousins at the extremes of the bell curve might share as little as 7 or 8 percent, or as much as 17 or 18 percent of their genome.
Why Two Cousins of the Same Degree Can Differ So Much
The variation comes from recombination, the process that shuffles chromosomes during the formation of eggs and sperm. Each time a parent passes DNA to a child, chromosomes swap segments at a handful of random locations. This means your mother did not hand you a perfect copy of either her mother’s or her father’s chromosome. She gave you a mosaic, and the pattern of that mosaic was determined by where the crossover events happened to fall.
DNA is inherited in long continuous stretches, not one gene at a time. In a pair of first cousins, the shared segments average about 25 centimorgans in length, which corresponds very roughly to 25 million base pairs. Because these inherited chunks are large and their boundaries are random, the total can vary quite a bit from one cousin pair to the next.
Over successive generations, recombination keeps breaking those segments into shorter pieces. A separation of about 50 meioses, roughly 25 generations back, produces shared segments averaging only about 2 centimorgans in length. That is small enough to be difficult or impossible for consumer DNA tests to detect reliably, which is why distant cousin matches often go unnoticed.
When Cousins Stop Sharing Any Detectable DNA
One of the most counterintuitive findings in genetic genealogy is that distant cousins frequently share zero DNA. Not a trace. Recent modeling confirms this in stark terms: about 98.5 percent of third cousins share at least one detectable segment, so nearly all of them show up as genetic relatives. But only about a third of fifth cousins share any segment at all. By eighth cousins, the figure drops below 1 percent.
This does not mean those eighth cousins are unrelated. They definitely descend from common ancestors. It means that recombination has broken the ancestral DNA into pieces too small to detect, or that by sheer chance, the segments that survived the shuffle were not passed to both individuals. Think of it like photocopying a document over and over: eventually the signal degrades past the point of recognition, even though the original existed. For genealogists hoping to confirm a family connection, this sets a practical ceiling. If your suspected relationship is more distant than about fourth or fifth cousins, a DNA test may simply come back blank.
Full Cousins Versus Half-Cousins
A full first cousin shares two grandparents with you, because your respective parents are full siblings. A half-first cousin shares only one grandparent, because your parents are half-siblings. The expected DNA sharing for a half-first cousin is about 6.25 percent, half that of a full first cousin. But the relationship between full and half-relatives is not just about the total amount of DNA. It also affects the pattern of sharing.
Full relatives share two lineages of common ancestry instead of one, which means they tend to inherit a larger number of individually smaller segments. Half-relatives of a comparable degree share fewer segments, but those segments tend to be somewhat longer on average. In practice, full relatives have a slightly higher probability of sharing at least one detectable segment compared to the half-relative of the same kinship degree. For example, about a third of full fifth cousins share at least one segment, compared to about 30 percent of half-fourth cousins once removed, even though those two relationship types involve similar total genetic distance.
How Consumer DNA Tests Measure Sharing
When a testing company tells you that you share 847 centimorgans with a match, it is reporting the total length of DNA segments you both inherited from a recent common ancestor. The key technical concept is identity by descent: two people share a stretch of DNA not just because the same letters happen to appear there (which can happen by coincidence in any two humans) but because they both inherited that exact stretch from the same ancestor.
Distinguishing genuine shared ancestry from coincidental similarity is a real challenge, especially at lower levels of relatedness. Modern tests use dense arrays of genetic markers, often more than half a million, to identify long continuous stretches where two people’s DNA matches. Short matches can be false positives, fragments that look shared by descent but are actually common in the broader population. Most testing platforms set a minimum segment length, typically around 5 to 7 centimorgans, below which they discard matches. This filtering is necessary to keep results accurate, but it also means some real but short segments between distant cousins go undetected.
Low-quality DNA samples introduce another layer of uncertainty. Research on forensic-quality specimens, such as DNA extracted from shed hairs, shows that genotyping errors and incomplete data can reduce the accuracy of kinship classification, particularly for close relatives. Adjusting the detection parameters, like the minimum number of matching markers and the tolerance for errors, can improve results without introducing false matches.
The X Chromosome Complicates Things
Most of the sharing percentages people discuss refer to the 22 pairs of autosomes, the non-sex chromosomes. The X chromosome follows its own inheritance rules, and those rules make cousin relationships asymmetric in ways that can surprise people.
Because men have one X chromosome (inherited from their mother) and women have two, the X does not recombine in males the way autosomes do. A father passes his single X intact to every daughter and no X to any son. This means the number of female ancestors along the lineage connecting two cousins is what determines how much X-chromosome DNA they might share. Two cousins linked entirely through a chain of male ancestors may share no X-chromosome DNA at all, while two cousins linked through a chain of female ancestors may share a disproportionately large amount.
For autosomal DNA, which makes up the vast majority of the genome, the sex of the connecting ancestors barely matters. But the X adds a subtle wrinkle. Genetic genealogists sometimes use X-chromosome sharing to narrow down which branch of a family tree a match comes from, since certain lineage paths make X sharing impossible.
Forensic Genealogy and Distant Cousin Matches
The practical limits of cousin DNA sharing have become a major factor in forensic investigations. Investigative genetic genealogy, the technique used to crack decades-old cold cases, relies on uploading crime-scene DNA to a genealogy database and finding distant relatives of the unknown suspect. Those matches are then used to build a family tree that converges on a single individual.
This approach typically works with relationships beyond first cousins, often in the range of second to fourth cousins, because closer relatives are less likely to appear in a public database by coincidence. The technique uses dense genetic marker data to infer distant relationships, and then conventional genealogy research fills in the family tree. The method has identified suspects in high-profile criminal cases and helped locate missing persons, though it raises significant privacy questions about the millions of people in these databases who never consented to forensic use of their DNA.
The drop-off in detectable sharing among distant cousins is a real constraint here. If the closest relative in the database is a fifth cousin, there is only about a one-in-three chance that a detectable DNA match even exists. Investigators often need multiple partial matches from different branches of the family tree to triangulate an identification.
When Cousins Marry and the Health Consequences
Shared DNA between cousins takes on a different significance in the context of consanguineous marriage, which means marriage between biological relatives. When two first cousins have a child, that child is predicted to have about 6.25 percent of their genome in a state where both copies of a gene are identical, inherited from the same ancestor. In practice, the observed figure tends to be higher. Research on children of first-cousin unions found that, on average, about 11 percent of their genome was homozygous, with a range of 5 to 20 percent across individuals. The excess above the theoretical 6.25 percent reflects the fact that in communities where cousin marriage is common, the parents themselves carry background homozygosity from earlier generations of intermarriage.
This elevated homozygosity increases the probability that a child inherits two copies of the same recessive mutation, which is the mechanism behind the higher rates of congenital heart conditions, kidney disease, and rare blood disorders documented in consanguineous populations. The average homozygous segment in children of first cousins was about 26 centimorgans long, large enough to encompass many genes. Globally, an estimated 10.4 percent of marriages are consanguineous, though the rate has been declining in most developed countries.
Why Your DNA Results Might Not Match Your Family Tree
One of the more common surprises from consumer DNA testing is a mismatch between what the family tree says and what the DNA shows. A person who believes they are first cousins with someone may discover a shared DNA amount that looks more like half-siblings, or more like second cousins. Several factors can explain these discrepancies.
The most straightforward is the natural variation described earlier. A pair of first cousins sharing 9 percent of their DNA is within the normal range, but testing companies may label that relationship as “second cousin” because their algorithms rely on population averages. The overlap in expected sharing between different relationship categories is substantial: a first cousin once removed, a half-first cousin, and a second cousin all hover in roughly the same centimorgan range, and algorithms cannot always tell them apart from DNA alone.
A more disruptive explanation is that the family tree itself is wrong. The widespread availability of consumer DNA tests has led many people to discover that a presumed father is not a biological father, sometimes called a non-paternity event. These discoveries can rewrite the expected cousin relationships across an entire family. If your father is not biologically related to his brother, then that brother’s children are not your first cousins at all, and the DNA will reflect it.
Endogamy and Small Populations
In populations with a long history of marrying within a closed community, the standard cousin-sharing expectations can break down in a different way. If your ancestors came from a small, endogamous group, you and a supposed second cousin might share substantially more DNA than the textbook 3.125 percent, because you are related through multiple recent ancestors rather than just one pair. This is not the same as cousin marriage in a single generation; it is the cumulative effect of a small mating pool over many generations.
For people with Ashkenazi Jewish, Icelandic, French-Canadian, or other historically endogamous backgrounds, this inflated sharing is well known in genetic genealogy communities. It means that the centimorgan values reported by testing companies can systematically overestimate how close a relationship is. Two people who share 200 centimorgans in an outbred population might be second cousins, but the same 200 centimorgans in an endogamous population could indicate a more distant connection supplemented by multiple shared lineages. Testing companies have begun to flag endogamous backgrounds in their results, but the adjustments are imperfect, and misidentified relationships remain common.
Historical research confirms that mate choice in many human populations was constrained both demographically and socially, ensuring that most unions involved partners who had inherited substantial DNA from common ancestors, even when close cousin marriage was formally discouraged. Community endogamy alone was often enough to produce high levels of genetic overlap across the group.
What Happens Beyond the Limits of Detection
If you go far enough back, every human alive today shares common ancestors, which means everyone is a cousin of some degree. The question is not whether you share ancestors with a random stranger but whether you share any detectable DNA from those ancestors. For most pairs of people from different continents, the common ancestors are so ancient that no identifiable segments survive the thousands of generations of recombination in between.
Within a single continent or ethnic group, the common ancestors are more recent, and the chance of sharing at least one small segment is higher. Studies of European populations have found that most pairs of individuals share at least a few very short segments traceable to ancestors within the last thousand years or so. But these segments are typically below the detection threshold of consumer tests, so they show up only in specialized research analyses that use lower filtering cutoffs.
Ancient DNA research has extended these methods into the deep past. Algorithms designed to detect shared segments in degraded archaeological specimens use probability models to distinguish real ancestral sharing from noise, allowing researchers to reconstruct family relationships in ancient burial sites and migration patterns across millennia. The same fundamental question applies whether you are comparing two living cousins or two skeletons from a Bronze Age cemetery: did these two individuals inherit the same stretch of DNA from the same ancestor, and is the segment long enough to prove it?