DNA testing for genealogy

A DNA test will not hand you a family tree. It gives you a list of relatives and a measure of how closely each one is related. This guide shows how to turn that list into answers.

The three kinds of DNA test

Test	What it follows	Who can take it	Best for
Autosomal	All your ancestral lines	Anyone	Finding relatives within about five to seven generations. Start here.
Y-DNA	Father's father's father, straight back	Males only	Surname questions and deep paternal lines
Mitochondrial	Mother's mother's mother, straight back	Anyone	Testing one specific maternal-line theory

Ethnicity percentages are estimates and they change when companies update their reference data. Your match list is the part that does genealogical work.

The size of the company's database matters more than anything else, because a bigger database holds more of your relatives. AncestryDNA is the largest. After you test, you can download your raw data file and upload it to other services, such as GEDmatch and FamilyTreeDNA, to find more matches without testing again. Check each company's current upload policy.

Reading shared DNA

Shared DNA is measured in centimorgans, written cM. More shared cM means a closer relationship. These are typical averages. Real values vary around them.

Relationship	Average shared DNA
Parent or child	about 3,400 cM
Full sibling	about 2,600 cM
Grandparent, aunt, uncle, half sibling	about 1,750 cM
First cousin	about 870 cM
First cousin once removed	about 430 cM
Second cousin	about 230 cM
Third cousin	about 75 cM

One number can fit several relationships. 1,750 cM could be a grandparent, an aunt or a half sibling. Use ages and known family facts to choose. The free Shared cM tool at DNA Painter lists every relationship a given number can represent.

Sorting your matches into family groups

You have four pairs of great-grandparents. Your second and third cousins each descend from one of those pairs. Sort the cousins into four groups and you have sorted your match list by family line. This approach is known as the Leeds method.

List your matches who share roughly 90 to 400 cM with you.
Take the first one and give it a color. Open the "shared matches" list for that person and give everyone on it the same color.
Go to the next match with no color. Give it a new color and repeat.
Continue until every match on the list has a color. Most people end up with about four groups.
For each group, find the surnames and places that repeat in the matches' trees. That tells you which great-grandparent pair the group belongs to.

A group you cannot connect to any known line is the most interesting one. It often belongs to the ancestor you are missing.

How investigators identify unknown people

Investigative genetic genealogy is the technique used to name unidentified remains and solve cold cases. Laboratories such as Othram, in Texas, specialize in the hard part: recovering a usable genetic profile from old, degraded or very small samples by sequencing the genome and reading hundreds of thousands of markers. The profile is then compared against genealogy databases whose users have agreed to that kind of matching.

What happens next is ordinary genealogy, done rigorously, and you can use the same steps on your own mystery:

Collect the strongest matches. Even third and fourth cousins are enough to start.
Build each match's tree backward until two or more matches meet at the same ancestral couple.
Find where the groups intersect. The unknown person descends from a marriage that joins two of those ancestral couples.
Build that couple's tree forward through children and grandchildren to list every candidate.
Eliminate candidates by age, sex, location and the amount of shared DNA predicted for each position.
Confirm with a direct test of a close relative before treating the answer as proven.

If you want your own DNA to help identify unknown persons, Othram runs DNASolves, a database built for that purpose. Contributing is your choice. Read its terms first.

Finding an unknown parent or grandparent

Test at the largest database, then upload your raw data elsewhere to widen the net.
Have a known relative from one side test too. Matches you share with them belong to that side. The rest belong to the unknown side.
Run the six steps above on the unknown-side matches.
When you have a candidate, look for supporting documents: was that person in the right place at the right time?

DNA results can reveal adoptions, donor conception and unexpected parents, for you and for the people you match. Decide how you will handle a surprise before you look. When you contact a new relative, be brief, be kind and never demand a reply.

Privacy before you test

Your DNA also describes your relatives, who did not consent. Think about that before uploading widely.
Each database has its own setting for law-enforcement matching. Read it and choose deliberately.
Use a strong, unique password and turn on two-step login for every DNA account.
You can delete your data and ask for your sample to be destroyed. Find out how before you send it.

Common questions

Which DNA test is best for genealogy?

An autosomal test from the company with the largest database is the best first test for almost everyone, because it finds the most relatives. Add a Y-DNA or mitochondrial test later only if you have a specific question about a direct paternal or maternal line.

Can I send my DNA to a forensic lab like Othram?

No. Forensic laboratories work on case evidence for law enforcement agencies and medical examiners. They do not process consumer genealogy kits. You can use the same matching and tree-building logic with a consumer test.

How many generations back does autosomal DNA reach?

It is reliable for about five to seven generations. Beyond that you inherit so little DNA from any single ancestor that many leave no detectable trace.

Next: Find the free records that turn a DNA match into a named ancestor.