Skip to main content
+31 85 109 1211 Beware of impersonators — verify our official details Verify official details

Blog

How to Recover a Missing Word in Your Seed Phrase (1–4 Words)

In short

Missing words in a BIP-39 recovery phrase can sometimes be reconstructed from the words and positions that remain. Each unknown position has 2,048 possible words; the checksum removes many combinations, but cannot identify the original wallet by itself.

  • Preserve the original backup and record known positions, partly legible letters, the phrase length and the wallet type. A known address or other reliable wallet data helps verify a candidate.
  • Each extra missing word greatly increases the work. One or two gaps are usually more manageable; no fixed count guarantees success or decides feasibility on its own.
  • Tools such as seedrecover.py can test candidates offline with the correct settings. Do not type partial or complete recovery phrases into websites. A phrase with no surviving words or other usable wallet material cannot be recreated from nothing.
How to Recover a Missing Word in Your Seed Phrase (1–4 Words)

There’s a gap in your recovery phrase. The paper tore, the ink faded, or you only ever wrote down eleven of the twelve words. Nothing about the gap changes what’s recorded on the blockchain: the coins stay put while you rebuild the way in. How realistic that rebuild is comes down to arithmetic, and the numbers are worth understanding before you try anything.

Why a gap can be filled at all

Every word in your phrase comes from a fixed list of 2,048 possibilities. That’s the entire reason a missing word in a seed phrase can be found: it isn’t an open-ended guess, it’s one of 2,048 known candidates per empty position. The phrase also carries a self-check woven into the words (how it works, and why wallets reject broken phrases, is explained in our invalid-phrase guide). Its job here is quieter: it silently disqualifies most candidates before anything gets tested. For a 12-word phrase with one blank, only about 128 of the 2,048 candidates even pass the self-check. For a 24-word phrase, about 8 do.

The last piece is one of your old receiving addresses. Candidates that pass the self-check each open a different wallet; the address is what identifies which of those wallets is yours.

Before you test anything

Write down what you still have, before your memory of it blurs: the surviving words in their positions, the phrase length, which positions are blank, and where the phrase came from. If the backup is torn or faded, preserve the original and avoid further handling. Keep the phrase offline; do not photograph it with a phone or camera that may sync images online. A partly legible word narrows a position from 2,048 candidates to a handful, and that fragment is worth protecting. Then find one old receiving address, from a transaction record, an exchange withdrawal, or an old wallet screenshot. A known address is useful, but not the only way to check candidates: an xpub or, for supported networks, a blockchain address database can also provide a reference.

If the phrase came from a Ledger or Trezor device, the brand-specific situations are covered separately: our Ledger seed guide and Trezor seed guide.

What the search costs, by number of missing words

One missing word: solved in seconds. 2,048 candidates, most eliminated by the self-check, the rest settled by your address. This is the best position to be in, and you can usually do it yourself with the tool below.

Two missing words: still fast. Roughly 4.2 million combinations. An ordinary computer runs through that in minutes when there’s an address to test against.

Three missing words: heavy but real. About 8.6 billion combinations. This moves past a laptop into GPU-based software that tests hundreds of thousands of combinations per second, and takes hours to days.

Four missing words: where a blind search gets expensive. Around 17.6 trillion raw combinations. Feasible with serious hardware and time, and any extra constraint (a partly legible word, a known first letter) cuts the space dramatically. Four is where we usually draw our own working line — a rule of thumb about cost, not a threshold that decides a case.

Five or more: rarely realistic. The space grows by a factor of 2,048 with every additional blank, so a search of that size is generally impractical without further constraints. Constraints can still move the picture — known positions, partly legible letters, a shorter phrase, a narrowed candidate set — which is why the count alone never settles a case. What no service can change is the arithmetic, and one that promises a result regardless isn’t being straight with you.

Recovering one missing word yourself

A seed-recovery tool does not search blindly. A BIP39 phrase has a checksum, and every word comes from the same 2,048-word list. That means a one-word gap is a bounded search: the tool combines the words you know with every candidate from the list, keeps only the combinations that pass the checksum, and verifies the result against one of your old receiving addresses. For a single missing word that is a fast, constrained search, not a brute force.

To run it yourself, use the seed-recovery mode of an open-source tool such as BTCRecover (the family seedrecover.py comes from). Download it from the official repository only, copy it to a computer you can disconnect, and work fully offline. You give it the partial phrase with the blank marked, the phrase length, the wallet type it came from, and one old receiving address - it tests the surviving candidates and returns the completed phrase when one matches. Never type a seed phrase, even a partial one, into any website: a phrase that touches an online form should be considered compromised, whole or not.

Restore the completed phrase directly in your wallet app, on your own device.

Several missing words, or positions you’re not sure of

BTCRecover handles multi-word gaps too, including the harder case where you have the words but don’t know which positions are empty. Be honest with yourself about the line, though: configuring GPU searches over billions of combinations is command-line work, and a wrong setup burns days without ever being wrong in an obvious way. If you’re missing two or more words and the tooling isn’t your world, this is the point where handing the case over makes sense rather than after a week of failed runs.

Practical limits and when reconstruction is impossible

With five or more missing words, a search is rarely realistic unless additional clues substantially narrow the candidates. Without a known address or other reliable verification data, there is nothing to confirm the result against: candidates that pass the self-check are all equally plausible, so even a small search can’t tell you which answer is yours. A completely lost phrase — no words, no address, no wallet file — can’t be reconstructed by anyone, and paying someone to try only adds a second loss to the first.

After the rebuild: close the gap for good

A phrase that lost words once will lose them again the same way. When access is back, move everything to a wallet with a freshly generated phrase and back that one up completely: all positions, numbered, count noted. Don’t continue on the patched original. What to write it on and where to keep it is covered in our backup-storage guide.

If words are missing from your phrase and you can point to one of your old addresses, Crypto Recovers will run the assessment for free and tell you honestly whether the search is realistic, before any commitment; the number of gaps is one input to that answer, not a qualifying threshold. No recovery, no fee, and the contract is signed before any work begins. The handover on success stays deliberately simple: you, the reopened wallet, and a new wallet only you control. You move the coins and settle the success fee yourself in one session, with both parties present. Your coins never sit with us at any point: not held, not stored, not managed.

FAQ: Missing seed word

Can a missing seed phrase word be recovered?

Usually, yes, when at least one address from the wallet is known. Every missing position is a search over the 2,048-word BIP39 list, and the checksum removes most candidates before testing. The harder the gap - more missing words, no address - the more the search costs, but a single missing word is normally solvable in minutes.

How long does it take to find a missing word?

One missing word: seconds to minutes on an ordinary computer. Two: minutes to hours. Three: hours to days with GPU-based software. Four: days to weeks with serious hardware and every additional constraint you can supply. The honest answer for your exact case depends on how many words are missing and what else you still have.

What if I don't have an old receiving address?

Without an address or other reliable verification data, the checksum alone cannot identify which candidate belongs to your wallet. An address from a transaction record, exchange withdrawal, or old wallet screenshot is the confirmation anchor - finding one is usually the most useful thing you can do.

Can the checksum alone recover a missing word?

The checksum narrows a single missing word from 2,048 candidates to a tiny set (about 128 for a 12-word phrase, about 8 for 24 words), but it can't tell you which candidate is yours. You still need an address or other reliable verification data to confirm the right one.

How do I recover a missing word from a Ledger or Trezor?

The method is the same as for any BIP39 wallet: mark the gap in the partial phrase, search with a tool like BTCRecover using one of your old receiving addresses, and restore the completed phrase on the device. The device itself doesn't need to be involved until you re-enter the full phrase. Ledger and Trezor specific situations are covered in our Ledger seed guide and Trezor seed guide.

I have all 12 words but not the order. Can the wallet still be recovered?

Usually, yes, as long as you have one old receiving address. Twelve known words in an unknown order give 479,001,600 possible orderings (12 × 11 × 10 … × 1, written 12!): far too many to try by hand, but a bounded job for software. The self-check discards most orderings before anything is tested, and your address confirms the right one. BTCRecover runs this offline as a token-list search (a GPU shortens it further), and every position you can pin down cuts the count sharply.

Someone may have seen 8 of my 12 words. Is my wallet still safe?

No. Eight known words leave only four unknown, which is 2,048⁴ (about 17.6 trillion) candidates: the same search this guide describes as feasible with serious hardware and time. An attacker doesn’t need your address, because every candidate can be checked directly against the blockchain. Move everything to a wallet with a freshly generated phrase now.

Still locked out?

The first assessment is free, and it asks for no wallet secrets. Describe what you still have — a file, a phrase, a device, or what you remember — and we’ll tell you whether recovery is realistic. You only pay if the recovery succeeds.

Request an assessment

Share this article

About the author

Robbert Bink

Robbert Bink

Founder & wallet recovery specialist

Completed his first successful wallet recovery in 2019. More than 15 years in programming and IT; leads case assessment, recovery strategy and specialist software development.

Read the full profile →