Understanding DNA matches for family history research

What a DNA match actually tells you

A DNA match is not a relationship. It is a piece of shared inherited material, and the relationship is something you still have to work out. When your results appear alongside a list of names, each with a predicted relationship and a shared centimorgan (cM) figure, it is tempting to treat that prediction as fact. Resist the temptation. The testing site is offering a statistical estimate based on how much DNA you share with that person compared with a large reference dataset. It is a very good clue, and nothing more.

What the match does tell you is that you and this person share at least one recent common ancestor. It does not tell you which ancestor, which line, or how many generations back. Two people can share identical amounts of DNA through entirely different routes. That is why every serious genetic genealogist pairs the match list with documentary research: birth, marriage and death certificates, census returns from 1841 to 1921, parish registers, wills and probate records, and the street directories and electoral rolls that fill the gaps between them.

Reading shared centimorgans sensibly

Centimorgans measure the length of shared DNA segments, and the figures cluster into recognisable bands. Roughly speaking:

• Parents and children share about 3,400–3,500 cM.
• Full siblings typically share 2,300–2,900 cM, though the range is wider than many people expect.
• Grandparents, half-siblings, aunts, uncles, nieces and nephews usually fall between about 1,150 and 2,000 cM.
• First cousins average around 850 cM, commonly between 550 and 1,300 cM.
• Second cousins average roughly 230 cM, with a broad range from about 45 to 515 cM.
• Third cousins average 73 cM, and the spread runs from zero to about 230 cM.

Notice how much those ranges overlap. A match at 400 cM could be a second cousin, a first cousin once removed, a half first cousin, or something more distant on a doubly related line. Treat any segment below about 7–10 cM with particular caution: small segments are frequently false, surviving unchanged across many generations by chance. Long segments are the reliable ones, which is why the chromosome browser matters more than the headline cM figure.

Clustering: turning a list into family groups

Most people begin with several hundred matches and no idea where to start. The answer is to sort them into groups before you try to name anyone. The standard approach is to look at your shared match lists: if a set of matches all share DNA with each other as well as with you, they almost certainly descend from one common ancestral couple. Build these clusters one at a time, giving each a neutral label such as "Cluster 1" until the paper trail names it.

Once you have clusters, work out which of your four grandparent lines each one belongs to. Test an older relative if you can — a parent, an aunt, or a great-aunt. Their results divide your matches immediately with far more clarity than your own, and testing the oldest available generation is the single most valuable thing most researchers can do. Then use the chromosome browser to check whether two matches share the same segment with you. When three people all share one identifiable segment, you have triangulation, which is much stronger evidence than any single match on its own.

When DNA and paper disagree

Sometimes a match is far closer, or far more distant, than your family tree predicts. This is not a reason to panic or to delete anyone from your tree. It is a reason to look again at the documents. Common explanations include an illegitimate birth, a second marriage that the family never discussed, a name change, or an adoption. Occasionally the explanation is simpler: an unrecorded deliberate estrangement, or a tree built on a coincidence of names in a crowded parish.

Also be alert to endogamy and pedigree collapse, where a community intermarried over generations or where cousins married cousins. In parts of rural Wales, the Scottish islands, and many immigrant communities, matches share far more DNA than their actual relationship suggests, and distant cousins can appear as second cousins. In those cases, cM totals are inflated and segment data is the only honest guide.

Two kinds of proof, not one

The strongest conclusions rest on both documentary and genetic evidence pointing at the same person. A shared segment tells you there is a biological connection; a certificate, a census entry or a will tells you who that connection was. Where the two agree, you can write the conclusion with confidence. Where only one exists, note it as a working hypothesis and keep looking.

• Record the shared cM figure and the date you saw it — match lists and estimates shift as more people test.
• Download your raw data and keep a copy; access to a testing site is not permanent.
• Note which chromosome and which segment overlaps, not just the total.
• Write your reasoning down as you go, so a future researcher can follow it.

Approaching matches politely

Finally, remember there is a person at the other end. Keep your first message short and warm, say who you are and what you are researching, and ask one clear question rather than presenting a research brief. Offer your own information freely. Never open with an accusation about a misattributed parentage, however confident you are, and give people time to reply. Some of the most rewarding moments in family history come from a stranger's willingness to share a photograph, a name, or a memory — and that cooperation is far easier to earn with courtesy than with certainty.