You tested, and something doesn't line up.

Before anything else, separate two situations that get conflated constantly and are not remotely the same problem. One is usually the technology behaving exactly as designed and being over-read. The other is harder evidence, and sometimes it means what you're afraid it means.

Situation one: the ethnicity estimate looks wrong

You have documented Irish ancestors and the result says four percent Ireland. Your family has said "we're German" for a century and Germany doesn't appear at all.

This is the more common complaint, and it is usually not a discovery about your family. It's a limitation of what an ethnicity estimate actually is.

It's an estimate against a reference panel. The company compares your DNA to groups of present-day people with deep documented roots in particular regions. Your result describes how your DNA compares to those panels — not where your ancestors lived. When a company improves its panels, everyone's percentages shift. People who have tested with the same company for years have watched their own results change repeatedly without their ancestry changing at all.

Regional borders are not genetic borders. Populations in neighboring regions of Europe are genetically very similar, because people moved, married, and were governed by borders that shifted every few generations. Distinguishing "German" from "Polish" from "Czech" DNA is not a clean problem, and the companies are honest about this in their documentation, which nobody reads.

And here's the one that resolves most cases: you do not inherit ancestry evenly.

You get half your DNA from each parent. But your parents each pass on a random half of what they have, so you do not get a clean quarter from each grandparent. It's roughly a quarter, and roughly can be off by a lot. Go back further and the variance compounds hard. By the fifth or sixth generation, it is entirely possible — and completely normal — to inherit no measurable DNA at all from a specific documented ancestor.

Read that again, because it defuses most of these situations. A great-great-great-grandmother can be thoroughly documented, absolutely real, and genetically invisible in your results. The paper trail isn't wrong. Her DNA just didn't make it down to you, and there's no mechanism that guarantees it would have.

Practical upshot: ethnicity percentages are the least reliable output of a DNA test. Treat them as a loose sketch. If your documented tree and your ethnicity estimate disagree, the documented tree — assuming it's sourced — is the stronger evidence.

Situation two: the matches don't fit

This is a different order of evidence entirely.

Match data — the list of people who share DNA with you, and how much — is hard information. The amount of shared DNA between two people falls into predictable ranges by relationship. It's not a percentage estimate against a reference panel; it's a direct measurement of shared segments.

So when a match doesn't fit the documented tree, something in the documentation is usually wrong.

The mundane causes, which are the majority:

  • An error copied from someone else's tree. Someone attached the wrong parents to a profile in 1998 and forty people copied it. This is the single most common cause of a tree that doesn't match its DNA.
  • A misattributed generation. A child raised by grandparents and recorded as their son or daughter. Extremely common historically, rarely written down as anything else.
  • A step-parent recorded as a parent. Remarriage after a death was routine, and records frequently don't distinguish.
  • A known adoption that simply stopped being discussed after a generation or two.
  • Two people with the same name in the same place, which is much more common than modern intuition suggests, particularly given naming patterns that repeat the same first names every generation.

And the cause people are actually afraid of: a biological parentage that differs from the documented one. In genealogy this is usually called a misattributed parentage event. It happens — through adoption, through affairs, through donor conception, through circumstances nobody recorded — and consumer DNA testing has made it visible at a scale that no previous generation experienced.

It is more common than most people assume. It is not rare, it is not shameful, and it is not something you did.

How to tell which you're looking at

Amount of shared DNA is the first question. Distant matches that don't fit are frequently just an error somewhere in a long chain of documentation, and chasing them is low-value. Close matches that don't fit — the ranges corresponding to parent, sibling, half-sibling, aunt, uncle, grandparent — are strong evidence and warrant careful attention.

Look at whether a whole side is missing. If you match your mother's side normally and match nobody on your father's documented side, that's a pattern rather than an anomaly.

Check the tree before the DNA. Genuinely — go back to the documented line and check what's actually behind each generation. A surprising number of these resolve into an unsourced link that someone guessed at decades ago and everyone inherited. Sources are the thing to check, one generation at a time.

Get a second opinion from people who do this. Genetic genealogy communities exist, they're populated by people who read shared-match data all day, and they will tell you within a few exchanges whether what you're looking at is an error or a finding. This is a technical skill and there's no reason to develop it alone at eleven at night.

What this article is not going to tell you

If it turns out to be situation two in its significant form, you're facing questions this page has no business answering. Whether to contact someone. Whether to tell your parents what you found, or your siblings. Whether you want to know more than you now know.

Those are yours. People who've been through it will tell you there's no schedule and no obligation to act on a result just because you have it — the information doesn't expire, and taking months before deciding anything is a completely normal response. There are communities and counselors who specialize in exactly this, and reaching one is a reasonable early step.

What the evidence is, and what the evidence isn't, is a technical question. What to do about it isn't, and anyone who conflates the two is overstepping.

The part that's easy to lose

A documented tree and a genetic tree are answering different questions, and both answers are real.

The people who raised you, the people whose names you carry, the family whose stories you know — a DNA result does not remove them from your history. It also doesn't invalidate the research. What it can do is add a line you didn't know about, and adding is not the same as replacing.

Most genealogists who encounter this end up maintaining both: the family that made them, and the biology that's now documented. Neither is a correction of the other.

Keep reading: