A shared cM number tells you how much DNA you and a match have in common, measured in centimorgans. On its own it rarely names the exact relationship. A total of 866 cM is the average for first cousins, but the same figure also sits comfortably inside the ranges for a great-aunt, a half-niece or a first cousin’s child. The number narrows the possibilities; your family tree, the ages involved and a few shared matches pick the answer.
Below you will find a range table built from published crowd-sourced data, a method for turning one number into a short list of candidates, and the traps that make a match look closer or more distant than it really is.
What a Centimorgan Measures (in Plain Terms)
A centimorgan is a unit of genetic distance, not a count of letters in your DNA. It describes how likely a stretch of a chromosome is to be split up when DNA passes from parent to child. Testing companies add up every stretch of DNA you share with someone that is long enough to count, and report the total in cM.
Two numbers usually appear on a match page:
- Total shared cM. The sum of all matching segments. This is the figure the range tables use.
- Number of segments, and the longest one. Close relatives tend to share a few long segments. Distant cousins share one or two short ones. A total spread across many small pieces can point to an older or more tangled connection.
Because DNA is shuffled at random in every generation, two people with the same relationship can share very different amounts. Full siblings are the classic example: they do not share exactly half of their DNA, only about half on average.
Shared cM Ranges by Relationship
The most widely used reference is the Shared cM Project tool at DNA Painter, which presents version 4.0 of Blaine Bettinger’s Shared cM Project (March 2020). The figures below come from that tool. Treat the range, not the average, as the useful part.
| Relationship | Average cM | Reported range |
|---|---|---|
| Parent or child | about 3,485 | 2,376 to 3,720 |
| Full sibling | 2,613 | 1,613 to 3,488 |
| Half sibling | 1,759 | 1,160 to 2,436 |
| Grandparent | 1,754 | 984 to 2,462 |
| Aunt or uncle | 1,741 | 1,201 to 2,282 |
| First cousin | 866 | 396 to 1,397 |
| First cousin once removed | 433 | 102 to 980 |
| Second cousin | 229 | 41 to 592 |
| Second cousin once removed | 122 | 14 to 353 |
| Third cousin | 73 | 0 to 234 |
| Fourth cousin | 35 | 0 to 139 |
Notice the overlap. Half siblings, grandparents and aunts or uncles all average around 1,750 cM. A match at 400 cM could be a first cousin at the low end, a first cousin once removed, or a second cousin at the high end. Notice also the zero at the bottom of the third and fourth cousin rows: some genuine third cousins share no detectable DNA at all.
From One Number to a Short List: A Four-Step Method
1. Write down every relationship whose range includes the figure
Do not stop at the first one. If a match shares 1,300 cM, your list includes grandparent, aunt or uncle, half sibling, great-grandparent, great-aunt or great-uncle, half-aunt, first cousin and a few more. The DNA Painter tool does this for you: enter the number and it lists every relationship it fits, with probabilities.
2. Cross out the ones that ages make impossible
A match born the same year as you is unlikely to be your great-uncle, though in large families it happens. Age gaps of 20 to 35 years suggest a generation difference. Many sites show an age bracket or let you message the match to ask.
3. Look at shared matches
The “shared matches” or “in common with” list shows people related to both of you. If the match shares DNA with your known paternal cousins but none of your maternal ones, you have halved the search at once.
4. Test each remaining candidate against your tree
Build the match’s tree a few generations back if they have none. Obituaries, census records and marriage notices are usually enough to find where two trees meet. The relationship that fits both the cM figure and the documents is your working answer, not the label the testing company printed.
A Worked Example With an Invented Match
This example is made up to show the method. Suppose a match called R. Ellis shares 312 cM across 14 segments, with a longest segment of 48 cM. R. Ellis is roughly your age and appears in your shared-match list alongside two known cousins from your mother’s father’s side.
- Candidates from the table: first cousin once removed, second cousin, half first cousin, second cousin once removed (high end), and a few others.
- Ages: similar ages make “once removed” less likely, though not impossible.
- Shared matches: points to your maternal grandfather’s family.
- Tree work: R. Ellis’s grandmother turns out to be a sister of your maternal grandfather. That makes the two of you second cousins, and 312 cM is a perfectly normal figure for that relationship.
The number did not prove the relationship. It ruled out most of the family and pointed the paper trail in the right direction.
When the Number Misleads You
The Shared cM Project tool states plainly that its statistics do not account for pedigree collapse or endogamy. Both can inflate shared DNA.
- Pedigree collapse. If two of your ancestors were themselves cousins, you descend from some people twice. A match from that family can share more than the relationship would suggest.
- Endogamy. In communities that married within a small group for many generations, everyone is related in several ways at once. Totals run high, and segment counts are often large with short segments.
- Double relationships. Two brothers who married two sisters produce children who are double first cousins, sharing far more than ordinary first cousins.
- Very distant matches. The tool notes that averages for distant relationships were calculated only from people who did share DNA. Below about 20 cM, a match may be a real but remote cousin or a coincidence of small segments, so treat those as low priority.
Keeping Track Without Losing Your Place
Once you have more than a dozen useful matches, a simple log saves you from re-checking the same people. One row per match is enough: name, shared cM, segment count, longest segment, which side of the family, the relationship you have proved and your source. Add a notes column for messages sent.
Contacting a match works best with a short, friendly note that gives your shared cM, the surnames you suspect connect you and one specific question. If writing to a stranger feels awkward, these ideas for reaching out after a long silence show how to keep a first message light and easy to answer.
For the wider picture of what the ethnicity estimate and match list can and cannot tell you, read our guide on what to do with DNA ancestry test results. And if the match leads you into a branch where you have no names yet, starting a family tree from scratch covers the free tools for building one.
Frequently Asked Questions
How many cM do first cousins share?
The Shared cM Project tool gives an average of 866 cM for first cousins, with a reported range of 396 to 1,397 cM. That range overlaps with first cousins once removed, half first cousins, great-aunts and several other relationships, so a total in this band needs ages, shared matches and a family tree before you can call it a first-cousin match.
Can you share DNA with a third cousin and not a second cousin?
Yes. Inheritance is random, and the amount passed down halves roughly each generation, with wide variation. Second cousins almost always share some DNA, but the low end of their range is around 41 cM. Third and fourth cousins can share none at all, so one cousin matching while their sibling does not is entirely normal.
Is a match under 20 cM worth researching?
It can be, but it should not come first. Small totals may reflect a real cousin many generations back or a chance match of short segments. Put your effort into matches above about 50 cM, where the trees are easier to connect, and return to the small ones when they share matches with a family line you are already working on.
Why does a different testing company show a different cM total?
Each company uses its own chip, its own minimum segment size and its own matching method, and some adjust totals to reduce matches from shared population background. Expect differences of several percent or more between sites. Compare like with like, and use the range table as a guide rather than a precise cut-off.