GEDCOM (GEnealogical Data COMmunication) is the file format family history travels in — a plain-text standard created by FamilySearch in 1984 that every genealogy program can read and write. A .ged file looks cryptic in Notepad, but the structure underneath is simple. Here it is.
A GEDCOM file is nothing but lines, and every line follows one grammar: a level number, an optional @-fenced ID, a tag, and a value. The level numbers create the hierarchy — a line at level 2 gives detail about the level-1 line above it, the way indentation works in an outline.
At level 0, a file is a list of records: a HEADer, then mostly INDI (individual) and FAM (family) records, plus supporting SOUR (source), NOTE, and OBJE (media) records, and finally a TRLR trailer that marks the file as complete.
The crucial idea: people are never linked to people directly. They are linked through families. An individual points at families with FAMC (“I am a child of this family”) and FAMS (“I am a partner in this family”); the family record points back with HUSB, WIFE, and one CHIL line per child. Siblings, then, are simply people whose FAMC points at the same family — the format never states sibling relationships at all. When these two-way links disagree, people silently vanish during imports, which is exactly what the validator checks for.
The exact form is 2 JUN 1985 — day, three-letter English month, year — but genealogy is full of uncertainty, so the format has qualifiers: ABT (about), BEF / AFT (before/after), EST and CAL (estimated/calculated), ranges like BET 1900 AND 1905, periods like FROM 1920 TO 1926, and dual years like 1750/51 for the old-style/new-style calendar switch. Partial dates — a year alone, or a month and year — are valid too.
Modern files are UTF-8, but GEDCOM predates Unicode, and files in the wild come in UTF-8, UTF-16, Windows-1252, and ANSEL — a pre-Unicode library standard with a famous quirk: accents are stored before the letter they sit on. A naive reader shows Göteborg as G¨oteborg. The declared 1 CHAR header line says which encoding a file uses, though it isn't always telling the truth — good readers verify rather than trust it.
| Version | Year | What it means in practice |
|---|---|---|
| GEDCOM 5.5 | 1996 | The long-reigning standard. ANSEL or ASCII text, 255-character lines, CONC/CONT continuation for long values. |
| GEDCOM 5.5.1 | 1999 / 2019 | A modest update — UTF-8 allowed, email/web fields, a few tags added and retired. A “draft” for twenty years yet the de-facto standard; it's what Ancestry, MyHeritage, and most desktop programs export today. Formally ratified in 2019. |
| GEDCOM 7.0 | 2021 | The real modernisation: UTF-8 only (no CHAR guessing), CONC abolished, no line-length limit, a clean extension mechanism, and GEDZIP — a zip that bundles the .ged with its photos, fixing the format's oldest complaint. FamilySearch's new tree export produces 7.0 natively. |
Is a .ged file the same as GEDCOM? Yes — .ged is just the file extension the GEDCOM standard uses. Does GEDCOM contain photos? Not in versions 5.x: it stores references to media files, not the files themselves (GEDZIP in 7.0 finally bundles them). Can I edit one by hand? It's plain text, so yes — carefully. Keep the level structure intact and re-check with the validator afterwards.
The fastest way to internalise the format is to read a small real file next to its rendered tree: download a sample GEDCOM file, open it in a text editor, and load the same file in the viewer — which shows every person's raw GEDCOM record alongside their profile.