Skip to content

Clean up after an import

After a large GEDCOM import there's almost always something to clean up: persons who came in twice, relationships that exist in several copies, and parents who exist as two different persons. The order you clean up in matters — if you do it in this order, you won't have to do the same thing twice.

Before you start

  • Make a backup. Merges can't be undone. → Backups
  • Did the whole import go wrong? Use Undo import in the import history instead of cleaning up by hand. → Undo an import
  • Everything imported has a tag, for example -gedcom-import-2026-11-02. Filter by it when you want to see exactly what came in.

1. Check the relationships

Go to Administration → Maintenance → Relationships → Check relationships and click Check.

Start with Fix automatically at the Partner and sibling relationships heading. It removes exact copies of relationships, creates missing directions, and fixes invalid subtypes — things that don't need your judgment. Source citations and notes on the copies are moved to the relationship that remains. Then click Remove copies under Exact copies of parent relationships.

Then go through Several different relationships between the same people. If the pair is both married and siblings, it's usually an import error — open the person by clicking the name and remove the relationship that's wrong. The back arrow takes you to the check again. If the relationships are correct, for example two marriages between the same persons, click Checked so the pair isn't shown again.

Wait with the Partner relationships and events group until step 5. After the merges, the couples are the right persons, and then what's shown there is accurate.

→ The relationship check in practice — common situations and what to do about them

2. Merge parents who exist twice

This is the best place to clean up families. The Children with more than one possible parent group shows each child with all of their parents, the parents' subtype, and years and places of birth and death. If the child has two fathers with the same name born the same year, it's almost always the same person imported twice.

  1. Read through the child's parents. Compare names, years, places, and subtype. Two mothers with almost the same name, born the same year or a year apart and in the same parish, are probably the same person. If one mother died before the child was born, or long before the other, it may instead be a first and a second wife — then one of the relationships is wrong, not a duplicate.
  2. Check the fathers — or mothers — who are the same person. You can check both fathers and mothers for the same child.
  3. Click Merge selected. The merge dialog opens one pair at a time, and the parent with the most details is preselected to be kept.
  4. Check the family context in the dialog — parents, siblings, and partners are at the same height for both — and merge.
  5. When all pairs are done, the check runs again. The child disappears from the list once it has only one father and one mother left.

Why here and not in the duplicate list?

The duplicate list shows two persons at a time. Here you see the whole family at once: the child, all the parents, and their details. That makes it easier to see who is the same person — and who really are two different persons, where one of the relationships is wrong.

If you don't have time for everything at once — click Tag all. The children get the tag -Check parents and are easy to find again.

3. Persons

Open Persons → Duplicates or Maintenance → Persons → Search duplicates.

  • Start with a high threshold, for example 80, and lower it once the certain pairs are done.
  • Click the name to open the person and take a closer look. The back arrow takes you to the same list, with the same tab and threshold.
  • Red warning about different dates? Check the source. It was common for a later child to be given the name of a sibling who had died, so two Annas with the same parents and different years of birth are often two persons.
  • Not the same person? Click Not duplicates so the pair doesn't come back.
  • Unsure? Write a note in the merge or tag the person -Investigate further.

→ Read more about duplicate handling

4. Places, and events last

Merge places before events, since the event duplicate search compares the places. Places are easiest to find with a low threshold, for example 30.

Do the events last. When two persons are merged, the one that remains gets both persons' events, so two marriages for the same couple often show up only after steps 2 and 3. Birth and death events with the same date and place are merged automatically when the persons are merged and aren't included in the event duplicates. The event duplicates show which persons are linked to the events, so you can see whether two marriages concern the same couple.

5. The relationships once more

Run the relationship check again. The merges often give new copies of relationships, and now it's time for Partner relationships and events: married couples without a marriage, dates that differ from the event, and events not linked to the couple's relationship.

Couples whose family had no marriage became partner in the import and got the tag -gedcom-partner-YYYY-MM-DD. Filter the person list by the tag and choose the right form for each couple.

Couples whose form came from a relationship status, for example a marriage with the type Unmarried, got the tag -gedcom-relationsstatus-YYYY-MM-DD. Couples living together in the past were usually married, so check the form for them too.

→ The relationship check in practice

6. What's left over

Under Maintenance:

  • Persons → No connections — persons without relationships, for example from a family that couldn't be imported
  • Places → Clean up unused — places that nothing points to anymore
  • Media → Clean up media — files that no media record points to

Finally, filter by the import's warning tag, for example -gedcom-warning-2026-11-02, to see records where the import found a problem.

See also