How to Find Unknown Relatives With DNA: A Practical Method

Finding unknown relatives with DNA comes down to five moves: test at a big company, upload your raw data free to the other databases, sort your matches into groups by which side of the family they're on, build trees outward from the matches you can identify, then confirm the connection with records. Most people find real living cousins within a few weeks of doing this properly.

The honest part first — DNA will not find every relative you have. Beyond about third cousin, sharing gets patchy, and plenty of genuine fourth and fifth cousins share no detectable DNA with you at all. What DNA does brilliantly is find the cousins who did test, and give you a way to place them. The free DNA match calculator will tell you what any shared centimorgan total could mean, which is the first thing you'll want when a stranger appears at the top of your list.

What Counts as an "Unknown Relative"?

This guide is about finding living cousins and lost branches of your family — the people your family lost touch with, the emigrant brother nobody heard from again, the great-aunt's descendants who ended up on another continent, or simply the cousins you never knew existed because nobody kept in touch after 1962.

That's a different job from searching for an unknown parent or a birth family, which needs its own approach and its own emotional preparation. The techniques overlap heavily, but the stakes and the etiquette are not the same.

Why DNA is so good at this

Paper genealogy naturally runs backwards. You start with yourself and work up through parents, grandparents, great-grandparents. It's brilliant at finding ancestors and terrible at finding the living.

DNA runs sideways. Your match list is a list of people alive right now who share ancestors with you. The database has already done the hard part — identifying that a relationship exists — and left you with the much smaller job of working out what it is.

Who Will DNA Actually Find? An Honest Table

Before you build hopes on it, know the shape of the limits. Two things decide whether a relative appears: whether they share enough DNA with you to be detected, and whether they've tested at all.

Two things follow from that table.

First, absence proves nothing. If a suspected fourth cousin doesn't appear on your match list, that isn't evidence you're unrelated. Recombination is random and some real cousins simply inherit no overlapping DNA from your common ancestors.

Second, the testing rate matters more than the biology. Most people have never taken a DNA test. Whole branches of your family may be entirely absent from every database — not because DNA failed, but because nobody in that branch spat in a tube.

What that means practically

Aim your effort at the second- and third-cousin range. Those matches are numerous enough to be useful, close enough to be confirmable, and their trees usually reach ancestors you can recognize. The distant matches are best treated as supporting evidence, not leads.

Step 1: Where Should You Test?

Test where the most people are. That's the whole strategy at this stage.

• AncestryDNA has the largest database by a clear margin, and it's the default answer for finding living relatives in the US, UK, Ireland, Canada and Australia. It has good tree integration and no chromosome browser.
• 23andMe has a substantial database and a genuinely useful DNA relatives feature, but weak tree tools and a lot of testers who came for health reports and never engage with matches.
• MyHeritage is comparatively strong in Europe, accepts free uploads, and includes automated clustering.
• FamilyTreeDNA is smaller, accepts uploads, and is the place for Y-DNA surname research and mitochondrial testing.

If you can only take one test, take the biggest. If you can take two, add one that's strong in the region your family came from.

Should you test relatives too?

Yes, and this is the most under-used move in genetic genealogy. Test the oldest generation you can reach.

Your parent carries roughly twice as much of your grandparents' DNA as you do. An aunt or uncle carries DNA you didn't inherit at all. Every older relative who tests reveals matches you'd never see through your own kit, and lets you sort your matches by which relatives they also match. One test of an 85-year-old great-aunt can be worth more than a year of your own research.

Step 2: Why Should You Upload Your Raw Data Elsewhere?

Because your matches are scattered across companies, and each database only shows you the people inside it.

Every major testing company lets you download your raw DNA file. Several others let you upload it, free or for a small unlock fee:

• MyHeritage — free upload, matches included; some advanced tools behind a small fee.
• FamilyTreeDNA — free upload, with extra features unlockable.
• GEDmatch — free upload; a third-party comparison site where kits from all companies meet.

The upload takes about ten minutes per site. You download one file, upload it three times, and multiply your reachable cousin pool.

The one caution

Read each site's privacy settings when you upload, and choose deliberately. GEDmatch in particular lets you opt in or out of law-enforcement matching, and that choice is yours to make consciously rather than by default. There's nothing sinister about any of this, but "I didn't realize" is a bad reason to end up anywhere.

Step 3: How Do You Read Your Match List?

Sort by shared centimorgans (cM), highest first, and start at the top.

Trust the number, not the label. Every site prints a relationship guess — "2nd–3rd cousin", "close family" — and every one of those is a probability, not a fact. A single cM total is consistent with several different relationships, because the amount of DNA shared by, say, a half-aunt and a first cousin once removed overlaps almost completely.

Here's the rough shape of it:

Those bands overlap heavily in real life, which is exactly why a calculator that shows you the full list of possibilities beats a single guessed label. The shared cM chart guide explains why the ranges are so wide, and our DNA matches explainer covers what all the other numbers on the page mean.

Start a research log immediately

Open a spreadsheet on day one. Name, shared cM, site, tree size, which side you think they're on, and what you've done about them. You will not remember any of this in three months, and re-deriving it is miserable.

Step 4: How Do You Work Out Which Side a Match Is On?

Use shared matches. This is the single most powerful free feature on every testing site.

Open any match and look at who else matches both of you. Those people belong to the same branch of your family, because the DNA you share came down the same line. Even if you don't know a single name in the group, you now know they belong together.

Cluster them. The Leeds Method is the standard way to do this by hand: list your matches in roughly the 90–400 cM range in a spreadsheet, then color-group anyone who appears on each other's shared-match lists. You'll usually finish with four groups, one for each of your grandparents' lines. It costs nothing and works at every company.

Once your clusters have names, every new match can be placed in seconds — you just check whose shared-match list they appear on.

If you already know some relatives

Even easier. If your known first cousin on your mother's side appears in a group of matches, that whole group is maternal. Every confirmed relative you can identify anchors an entire branch.

Step 5: How Do You Build Trees Outward From a Match?

This is where most people stall, and it's the step that actually finds people.

The instinct is to look at your own tree and search for a connection. That rarely works, because your unknown match usually connects through a branch you know nothing about.

The technique that does work runs the other way:

1. Pick a match with a readable tree — public, several generations deep, in a cluster you care about.
2. Work up their tree to the generation where you'd expect a common ancestor. For a second cousin that's great-grandparents; for a third cousin, great-great-grandparents.
3. Then work back down, tracing all the descendants of that couple, not just the line leading to your match. Every child, every grandchild.
4. Look for the point where the descendants approach your own family — a surname, a town, a date that connects.

You're building a tree of an entire family and seeing where you fit into it. Genealogists call the tree-of-a-whole-family approach a descendancy tree, and it's the workhorse of this kind of research.

Where to do the tree work free

You don't need a subscription to make progress. FamilySearch is free and enormous, with a shared global tree and huge record collections. Our guide to tracing your family tree free covers the rest of the no-cost options — free record sites, indexes, and where to look when the obvious search returns nothing.

For living and recently deceased people, obituaries are the single best resource. An obituary typically lists surviving siblings, children and grandchildren by name, which is exactly the map you need to connect a tree to a living match.

Step 6: How Do You Confirm the Relationship?

A plausible tree and a matching cM total is a hypothesis, not a fact. Confirmation means three things lining up:

• The DNA fits. The shared amount is within the expected range for the relationship you're proposing.
• The paper fits. Records document each link in the chain — births, marriages, censuses, deaths.
• The other matches fit. Other people in the same cluster also connect to the same ancestral couple, at amounts consistent with their positions.

That third one is the strongest test. A single match can be a coincidence, especially at lower cM levels. A whole cluster of matches all descending from the same couple is very hard to explain any other way. Our guide on confirming a DNA match walks through what solid proof looks like and the traps that produce false confidence.

Beware the plausible-but-wrong answer

The classic error is finding a connection that works and stopping. If your match's tree contains a surname from your own tree, it's tempting to declare victory. But surnames repeat, families intermarried, and in some communities everyone is related several ways at once. Check whether an alternative explanation also fits before you settle.

How Should You Message a New Relative?

Short, warm, low-pressure. Most people over-write their first message by a factor of ten.

A good first message:

Hi — we show up as DNA matches (about 180 cM). I think we might connect through the Brennan family from County Mayo. Does that name mean anything to you? No worries if not, and thanks either way.

That's it. Three sentences. It says who you are, gives them one concrete thing to react to, asks one easy question, and explicitly gives them permission to say no.

What to avoid:

• Life-changing news out of the blue. If your research suggests something significant about their parentage or family, that does not belong in a first message from a stranger. Build a relationship first, or don't send it at all.
• The full research dossier. Nobody reads eight paragraphs from someone they've never met.
• Demands. "Please send me your tree / your GEDCOM / your grandmother's maiden name" reads as extraction, not connection.
• Multiple messages in a week. One message. If nothing comes back, one polite follow-up months later at most.
• Anything about money. Ever.

Follow their lead

If someone replies briefly, reply briefly. If they want to talk for an hour about their grandmother, wonderful. Some matches become genuine friends and some want one factual exchange and nothing more. Both are fine, and you don't get to choose which.

Respect non-responders completely

Silence is an answer. Don't find their Facebook, don't email their workplace, don't message their children. The line between enthusiastic genealogist and someone being harassed by a stranger is thinner than it feels from the inside, and it's the one rule of this hobby worth being strict about.

What If Nobody Replies?

It happens constantly, and it usually isn't about you.

The majority of DNA testers bought a kit to see an ethnicity pie chart. They looked at it once, never opened the match list, and haven't logged in for three years. Many don't even have notifications turned on.

What still works when messages don't:

• Their tree may be public even if they're silent. All the research value, none of the conversation.
• Their shared matches still cluster. A non-responding match is still a useful data point for placing everyone around them.
• Other people in the same cluster may reply. You don't need any specific person — you need one person from that branch.
• New matches arrive constantly. The cousin who answers everything may test next year.
• Public records don't need permission. Obituaries, censuses and newspapers will often tell you what a silent match could have told you. Old newspaper archives are especially good for connecting families to places.

What If You Find Something You Weren't Expecting?

Sometimes a search for cousins turns up something else — a close match nobody can explain, a cluster that doesn't correspond to any family you know, a half sibling.

Slow down. There's no deadline. DNA is telling you something factual about biology, and absolutely nothing about who raised whom, who loved whom, or what any of it meant to the people involved.

Before contacting anyone, sit with it for a while. Consider who else the information affects — a living parent, a sibling, an elderly relative who may have kept a secret for fifty years for reasons that made sense to them. Genetic genealogy support communities exist precisely for this moment, and talking to someone who's been through it helps more than another evening of research.

And if you're the one who receives an unexpected message: you owe nobody a relationship, a reply, or an explanation.

FAQ

Can DNA testing find living relatives?

Yes — that's what a match list is. Every person on it is alive (or their kit is managed by someone alive) and shares ancestors with you. What DNA can't do is find relatives who haven't tested, or those distant enough that you share no detectable DNA at all.

Which DNA test finds the most relatives?

AncestryDNA has the largest database, so it typically returns the most matches, especially for people with US, UK, Irish, Canadian or Australian roots. Testing there and then uploading your raw data free to MyHeritage, FamilyTreeDNA and GEDmatch gives you the widest possible reach for one test fee.

Why can't I find a cousin I know exists?

Either they haven't tested, or you genuinely share no detectable DNA. Beyond third cousin, sharing becomes patchy — commonly cited estimates put fourth cousins near a coin flip, and fifth cousins much lower. Not appearing on your match list is not evidence that two people are unrelated.

How do I know which side of my family a match is on?

Use shared matches. Anyone who matches both you and a known relative belongs to that relative's side. If you know nobody, use the Leeds Method to color-group your matches into four clusters — one per grandparent line — then identify each cluster from the trees of the people inside it.

What should I say in a first message to a DNA match?

Keep it to two or three sentences. Say you're a DNA match, mention the shared cM and any surname or place you suspect connects you, ask one easy question, and make it clear a non-reply is fine. Save detailed research for after they've written back.

Is it okay to contact a DNA match who hasn't replied?

One polite follow-up after a few months is reasonable. Beyond that, stop. Don't pursue them on other platforms, contact their family members, or reach them through work. Many testers never check their accounts, and a non-reply should be treated as a closed door.

Start With the Numbers, Then Start Knocking

Finding unknown relatives isn't one clever trick — it's a stack of ordinary steps done in order. Test widely, upload everywhere, cluster your matches, build trees outward, confirm with records, and write short kind messages. Do that consistently and cousins turn up.

Begin with the match sitting at the top of your list right now. Run their shared cM total through the free DNA match calculator to see every relationship it could represent, then read the shared cM chart guide so the overlapping ranges stop being confusing. Once you can read the numbers, the rest is just patient, ordinary genealogy.