Clear Kindle Vocabulary Builder:
there is no delete all.

Nov 3, 2026 · 5 min read

Kindle has no button that clears Vocabulary Builder. You remove words one at a time, with Delete or Mastered, and reports going back to 2014 say deleted words can return because the list syncs with your Amazon account. Export the list first, then delete.

Clearing it from the Kindle

You cannot clear Vocabulary Builder in one go from the Kindle's own menus. MakeUseOf's walkthrough (November 2022) does not mention a clear-all option, and Nerdschalk (updated April 2024) describes only the same per-word route: open the word, tap Delete.

The per-word route is slow for a list of hundreds, and the list is tied to your Amazon account. So the useful order is: save a copy of what you have, delete what you do not want, then check the words have stayed gone.

Save the list before you delete anything

Deleting is the part you cannot take back, so do the export first. Vocabulary Builder keeps your words in a file called vocab.db on the Kindle, and Amazon has never given it an export button. The free converter at /kindle-vocab-to-csv turns that file into a CSV or an Anki deck in your browser, and nothing is uploaded. The longer walkthrough, including a one-line Terminal version, is on the export page.

Whatever you use, copy vocab.db off the Kindle and keep that copy untouched. It is the only backup you will have.

Delete words on the Kindle, one by one

MakeUseOf's steps: tap the top of the screen in a book, tap the three dots at the top right, choose Vocabulary Builder, then tap a word. At the bottom you get two choices. Delete removes the word. Mastered hides it for now, and the Kindle brings it back later for self-testing.

That difference matters if the goal is a clean slate. Mastered words are not gone, they are filed under a different label. In the database, the WORDS table has a category column, and in the 2014 MobileRead thread below, 0 is a word you are still learning and 100 is a mastered one. Menus differ between firmware versions, so if yours does not match, look for the same three dots menu in a book.

Why deleted words come back

The list syncs to the cloud. In a MobileRead thread called "Vocabulary builder reset", users tried to wipe the whole thing and ran into this several times. In April 2014 a user reported that Amazon support said the data lives on their servers, so words return after a factory reset and a Wi-Fi connection. In December 2014 another user emptied all four tables and wrote that after a restart the Kindle "keeps downloading the old db from amazon servers".

The same thread has a more hopeful report. A poster who goes by dtp wrote that deleting the file just makes the server send it back, while removing entries inside the database is treated as you deleting the words yourself. Another user confirmed in April 2014 that the entry-deletion method worked. A third, Tx65, said that after deleting rows you also have to compact the database before the list shows as empty. A MobileRead poster also suggested doing the deletion in Airplane mode, and was upfront about not having tried it.

All of this is from 2014. No current report either way could be found, so treat it as a description of how the feature behaved then. The practical test is cheap: delete a few words, restart the Kindle, reconnect to Wi-Fi, and see whether they return.

Editing vocab.db yourself, on a copy

For people comfortable in Terminal. Work on a copy on your Mac, never on the file on the Kindle's drive, and keep the untouched original from the export step. No Kindle was available, so these statements were tested with the macOS sqlite3 tool (version 3.51.0) on a scratch database built from the schema documented by readers: WORDS (id, word, stem, lang, category, timestamp), LOOKUPS (word_key, book_key, usage, timestamp) and BOOK_INFO (id, title, authors). On that scratch file they did what is described below. A real vocab.db has at least one more table, DICT_INFO, and probably extra columns, so run .schema on yours first and compare.

To empty everything, then compact the file:

sqlite3 vocab-copy.db "DELETE FROM LOOKUPS; DELETE FROM WORDS; DELETE FROM BOOK_INFO; VACUUM;"

To remove only the words you are still learning (category 0) and keep the mastered ones, delete the lookups that belong to those words first:

sqlite3 vocab-copy.db "DELETE FROM LOOKUPS WHERE word_key IN (SELECT id FROM WORDS WHERE category=0); DELETE FROM WORDS WHERE category=0; VACUUM;"

The order matters, because LOOKUPS points at WORDS through word_key. The thread shows what happens when you run the wrong line: in May 2014 a user ran only the learning-words and lookup deletes and found the mastered words gone as well. Doitsu replied that the wrong line had "most likely" been executed, and that you should back the file up first. Test on the copy, count the rows afterwards with SELECT count(*) FROM WORDS;, and only then think about putting anything back on the device.

Two warnings. Putting an edited file back on the Kindle is exactly the move the 2014 reports describe as sometimes working, and no report with current firmware could be found. And anything you change on the device can be overwritten when it next syncs with Amazon.

The jailbreak route, and why to skip it

In the same MobileRead thread, a user named Doitsu posted an extension for KUAL, the launcher used on jailbroken Kindles, in August 2014. Its menu includes Delete all, which wipes WORDS, LOOKUPS, DICT_INFO and BOOK_INFO, plus an Export all that writes the words and lookups out as text files. Doitsu's own warning was that none of its queries are reversible and that you should back up vocab.db first.

That means a jailbreak, a 2014 script and an irreversible command on a device that syncs to your account. If all you want is a shorter list, deleting by hand is safer. This route is only worth a look for an old Kindle you do not mind losing the list on.

If the list is empty or the option has vanished

A different problem with the same symptom. In a January 2024 MobileRead thread, a user on a Paperwhite 11th generation running firmware 5.16.4 could not find the Vocabulary Builder option. Another poster guessed that registration was required because the list is stored in the cloud, and the original poster later wrote that registering the Kindle was what fixed it. Check the Kindle is signed in to your Amazon account before you assume anything is broken.

Going forward, words you look up on a Mac are a separate matter. LeafyApp saves them in a library on your Mac with the sentence, source and date. It does not read vocab.db, so it will not touch your Kindle list either way, and it does not sync with it.

Questions people ask

How do I clear Vocabulary Builder on a Kindle?
There is no clear-all button. Open Vocabulary Builder from the three dots menu in a book, tap a word, and choose Delete at the bottom, one word at a time. Export the list first with the free vocab.db converter.

Can I delete all Vocabulary Builder words at once?
Not from the Kindle's menus. Older community reports describe emptying the database tables on a computer, or a KUAL extension on jailbroken Kindles, and also report that the list can sync back from Amazon. Both are from 2014, so test on a copy and expect surprises.

How do I reset Vocabulary Builder on a Kindle?
There is no reset option in the menus in any source found. A 2014 MobileRead thread reports that a factory reset did not help, because Amazon's servers sent the words back once the Kindle reconnected.

Why do my deleted Kindle words come back?
Because the list is stored with your Amazon account and syncs to the device. Users in a 2014 MobileRead thread reported old data being downloaded again after deleting the database file or emptying its tables.

What is the difference between Delete and Mastered?
Delete removes the word. Mastered hides it from the active list, and the Kindle brings it back later for self-testing, according to MakeUseOf (November 2022).

Kindle and Amazon are trademarks of Amazon.com, Inc. or its affiliates. LeafyApp is an independent app, not affiliated with or endorsed by Amazon, and the names are used here only to describe compatibility.