How to Split a Large Family Tree Before Uploading
Splitting a GEDCOM file before uploading is the fix genealogists repeat most often on FamilySearch's own forum. Here's how to do it without breaking links.
If you've read why FamilySearch uploads keep failing, you already know the fix genealogists repeat most often in FamilySearch's own community forum: split the file into smaller pieces before uploading. This guide covers how to actually do that without hand-editing raw GEDCOM text.
Why splitting works
FamilySearch's documented 100 MB file-size cap is a hard wall, and files with more than roughly 1,000 individuals have been repeatedly reported to lose data silently during upload — the process "completes" but only part of the tree shows up. Both problems shrink the same way: fewer individuals in the file means a smaller file and less structure for the importer to choke on. Splitting doesn't fix a broken file; it reduces the odds that a large, otherwise-valid file trips a size or complexity problem you can't see coming.
The three practical ways to split
Direct ancestors of a person. Pulls out everyone in that person's ancestral line — parents, grandparents, and so on back through every FAMC (family-as-child) link — without including siblings, cousins, or descendants. This is the right choice if you're uploading your own line and don't need collateral relatives.
Descendants of a person. The reverse: everyone descended from a chosen ancestor, following FAMS (family-as-spouse) links forward through every generation, including spouses who married in. Useful for sharing one branch of a large tree with a specific relative.
Everyone born after a given year. A flat filter rather than a relationship walk — useful when your tree spans many generations and you just need to cut its overall size, regardless of which branch each person belongs to.
Doing it with the Tree Splitter
The Tree Splitter runs all three of the above entirely in your browser — your file is never uploaded anywhere to be split. Drop in your GEDCOM, pick a mode, and for the two relationship-based modes, search for the person by name to use as your starting point. For the birth-year mode, just set a year.
The part that matters most happens after you click extract: the tool doesn't just delete the people you didn't select. It also identifies every supporting record those remaining people reference — sources, repositories, media objects, notes, submitters — and pulls those in too, so citations and attached media survive the split. Then it runs one more pass that strips any pointer left dangling by the extraction (a reference to a person or family that got excluded), because a subset file with broken references would recreate the exact "dangling reference" problem this whole product exists to help you avoid. You get a plain, valid GEDCOM file back — smaller, but structurally complete.
After you split
Run the result through the Upload Readiness Checker again before uploading. If it's still flagged, split further — the tool works on any GEDCOM file, including one you already extracted, so narrowing to a smaller ancestor, descendant, or year range is just repeating the same steps on the smaller file.
If you specifically want your own direct line rather than a general subset, see Exporting Just Your Direct Ancestors: A Step-by-Step Guide for the exact walkthrough.
Do it now
Keep reading
Common questions
- Won't splitting my tree lose data?
- Not if the tool does it correctly. The Tree Splitter strips every pointer that would reference someone left out of the subset, so the exported file has no dangling references — it's a smaller, complete file, not a corrupted one.
- How do I decide which split to use?
- If you're building your own line back in time, use direct ancestors. If you're sharing one branch with a relative, use descendants of a shared ancestor. If your file just has too many people spread across generations, splitting by birth year is often the fastest way to get under FamilySearch's recommended range.
- Do I need to split more than once?
- Possibly, for very large files. Run the Upload Readiness Checker on each resulting piece — if a subset is still flagged as high risk, split it again by a narrower ancestor, descendant, or year range.