Sample GEDCOM files are the fastest way to see what a family tree file actually contains. Whether you're testing a viewer, learning the format, or just curious what 3,000 European royals look like in plain text, a good sample file beats reading the specification. All five below are free, publicly available, and safe to open.
You can drop any of them straight into the GEDCOM viewer to browse the tree without installing genealogy software, or run them through the converters to see how they translate into PDF, Excel, HTML, JSON, or XML.
1. royal92.ged — 3,010 European royals from 1992
This is the most-copied sample GEDCOM file in existence, and it has aged into a genuine artifact. Denis R. Reid compiled it in Personal Ancestral File 2.2 and released it publicly in November 1992, wrapping the announcement in a Christmas greeting that is still sitting in the file header today.
- Size: 3,010 individuals, 1,422 families
- Timeframe: 534 to 1992
- Format: GEDCOM 5.5, ANSEL encoding
- Download: D-Jeffrey/gedcom-samples
The very first record is Queen Victoria, born 24 May 1819 at Kensington Palace. From there you can walk backwards through most of the crowned heads of Europe.
Two things make it useful beyond the novelty. First, the ANSEL encoding is a great test of whether your software handles pre-Unicode character sets properly. Second, the data has known errors. Reid himself notes in the header that the French and Spanish lines are the weakest part, and Victoria is labelled “Queen of England” rather than Queen of the United Kingdom. That makes it a realistic sample rather than a sanitised one.
2. Harry Potter.ged — a fictional tree with real quirks
Findmypast maintains a small public repository of test files, and the Harry Potter tree is the standout. It was exported from Family Tree Maker in December 2007 and covers the Potter, Black, Weasley, and Dursley families.
- Size: roughly 16 KB, around 180 individuals across 31 families
- Format: GEDCOM 5.5, ANSI encoding
- Download: findmypast/gedcom-samples
The custom event entries are the fun part. You'll find things like “Gave gold to his runaway nephew” and “Married Muggle Tom Tonks” sitting in EVEN tags, which is a nice illustration of how flexible GEDCOM's event structure is.
There's a technical bonus too. The file declares ANSI encoding, which was never a legal value in GEDCOM 5.5 (only ASCII, ANSEL, and Unicode were). Strict parsers will complain. That makes it an accidental but effective test of how gracefully your tools handle malformed headers — run it through the GEDCOM validator and the header problem is the first thing reported.
One note: the findmypast repository references a private vault for files containing real personal data. Stick to the clearly public files like this one.
3. bach.ged — 33 people, five generations of composers
Not every sample GEDCOM file needs to be enormous. The Bach musical dynasty file is small enough to read end to end in a text editor while you follow along with the spec, which makes it the best learning file on this list.
- Size: 33 individuals, 8 surnames
- Timeframe: 1550 to 1871
- Format: GEDCOM 5.5, exported from PAF 2.2
- Download: D-Jeffrey/gedcom-samples
Johann Sebastian was one node in a family that produced working musicians for seven generations, and seeing it laid out as a tree makes the scale of that obvious in a way a Wikipedia article doesn't.
If you want more in this vein, the same repository has shakespeare.ged (31 people, tracing William back to his grandfather Richard) and bronte.ged (14 people, the writing sisters and their father Patrick). All three are historical, tiny, and free of any living-person concerns.
4. maximal70.ged — the official GEDCOM 7 test file
Everything above is GEDCOM 5.5. If you want to see where the format is now, this is the file. FamilySearch publishes it at gedcom.io as part of the official GEDCOM 7.0 test suite, and it exercises every standard tag and every standard enumeration value in the specification.
- Size: 700+ lines
- Format: GEDCOM 7.0, UTF-8 with byte order mark
- Licence: Apache 2.0
- Download: gedcom.io/testfiles/gedcom70
It's a deliberate stress test rather than a real family, and the header says as much: the data isn't meaningful genealogically. What it does is break assumptions. The first top-level record is a FAM, not an INDI, which trips up any parser that expects individuals to come first. It uses @VOID@ pointers, custom schema tags, and LDS ordinance records — the last of which have a long history of their own.
Pair it with minimal70.ged from the same directory, which is the smallest legal GEDCOM 7 file you can write — essentially a header, a version declaration, and a trailer. Seeing the floor and the ceiling side by side teaches the format quickly. There are also .gdz versions of both, demonstrating the new GEDZIP container that bundles media alongside the data.
5. Married1200.ged — one man, 1,200 wives
Tamura Jones hand-built this file in 2010 for an article about the outer limits of what genealogy software will accept. It contains exactly what the name says: a single individual linked to 1,200 spouses.
- Format: GEDCOM 5.5, declared as ANSEL but using only ASCII characters
- Download: tamurajones.net
It's not a pleasant browse. It is an excellent way to find out whether a program renders a person's spouse list, paginates it, or simply gives up. Jones also published Children1200.ged and a pair of long-name tests that check how software copes with unusually long lines.
If you want more in this category, the classic GEDCOM 5.5 torture test files (TGC55 and its variants) are still hosted at geditcom.com. They use every legal tag in every legal position, and they've been breaking genealogy software for over twenty years.
Honourable mentions
- pres2020.ged — 2,322 individuals covering every US president and their ancestors. Genuinely interesting, but it flags over 1,000 people as living, so treat it as a private test file rather than something to republish.
- Greek Gods.ged — mythology as genealogy, from the Famous Family Trees project on SourceForge. There are Norse, Roman, and Biblical equivalents.
- Queen.ged — 4,683 people with dates reaching back to 4004 BC, which makes it the go-to file for testing BCE date handling.
A note on privacy
GEDCOM has no meaningful privacy model. Anything in the file is fully visible to anyone who opens it. Several widely circulated sample files contain hundreds or thousands of people flagged as living, including the presidential and Habsburg trees. For anything you plan to share publicly, stick to the historical files (royal92, Bach, Shakespeare) or the fictional ones.
The same caution applies to your own tree. Every tool on this site parses files in the browser and uploads nothing — so opening a file with living relatives in it, converting it to a shareable web page, or checking it with the validator never sends the data anywhere. What you then hand to other people is the part worth thinking about.
Start with a file, not the specification
Every file listed here is free to download. Grab one, open it in the viewer, and see how your own tree compares. If you haven't exported yours yet, the export guides cover Ancestry, MyHeritage, FamilySearch, and Family Tree Maker — and if you have family data sitting in a spreadsheet instead, the Excel to GEDCOM and CSV to GEDCOM converters will turn it into a valid .ged file to test with.