You should not need to read this if your backup has the normal 12 or 24 BIP39 words and you know exactly what they are. But sometimes you open your hardware wallet, type in the seed phrase from your backup, and realize it is too long. Maybe you added extra words years ago because you thought that would make the seed safer. Maybe you cannot remember which words were extra.
If that has happened to you, this guide explains what is going on, why adding valid BIP39 words does not really protect your seed, and how I use BTCRecover to recover from a backup that contains too many seed words. Watch the video if you want to see the same process on screen.
Stuck with a wallet recovery?
I maintain BTCRecover and have helped people recover seed phrases, passphrases and wallet passwords since 2019. Book a one-hour video session and I’ll walk you through doing the recovery yourself, without sharing your seed. Or, if you’d rather I run it, a trusted recovery is charged only if it succeeds.
Private sessions: US$100 per hour. Trusted recovery: 10% of recovered funds. I never contact people first and will never ask for your seed phrase over chat or email.
Why extra BIP39 words do not make your seed much safer
The first thing I want to be clear about is this: putting extra valid BIP39 words into a mnemonic does not give you the kind of security people think it gives them.
A lot of people imagine that if they add ten random-looking words, an attacker would have no idea which words belong there and which ones do not. In practice, that assumption is weak. If the extra words are valid BIP39 words, a recovery tool can treat each unknown word as something like a large typo and try removing it from different positions in the phrase.
That means if you only added a small number of extra words, an attacker may be able to brute-force their removal very quickly. It might feel safer when you are writing down the backup, but it is not a strong security improvement. If your goal is to add an extra layer of protection to a BIP39 seed, a BIP39 passphrase is usually a much better option than inserting extra mnemonic words.
The situation this guide covers
The example in the video is a best-guess seed that is 34 words long instead of 24. In other words, there are ten unknown extra words somewhere inside the phrase. All of those extra words happen to be valid BIP39 words, which makes them harder for a human to identify by eye.
The goal is simple: recover the correct 24-word mnemonic from a longer backup that contains too many seed words.
This kind of recovery can also apply when you have added only a few extra words. The more unknown extra words you need to remove, though, the longer the search takes. That part matters a lot, so I will cover it later in this article.
Important security warning before you start
Before you run any recovery tool against your own seed phrase, think carefully about the computer you are using.
BTCRecover is not designed to keep your seed secret from other processes running on the same machine. If you are working with a live seed that controls real funds, I strongly recommend doing this only after you have practiced with example seeds first. For your own recovery, an air-gapped or otherwise trusted environment is much safer.
If you are not comfortable setting up a safe recovery environment yet, do not jump straight into testing your real backup on a normal internet-connected desktop. Practice the commands with examples until you understand what each option does and how the tool behaves.
Get BTCRecover ready
You will need to download BTCRecover from GitHub. In my videos I usually recommend using my fork of BTCRecover because it has extended functionality, including support for coins beyond Bitcoin, and ongoing updates toward Python 3. If you have followed my earlier BTCRecover setup guides, this part should feel familiar.
The tool runs in Python, so it can be used on Windows, Mac, or Linux. The exact setup steps depend on your platform, but the recovery command-line approach is similar across systems.

Practice with example seeds first
If this is your first time using BTCRecover for this kind of problem, I recommend reproducing the examples from the video before touching your own seed. The commands, example addresses, and other details are in the video description.
Practicing on examples gives you a few important benefits:
- You learn how to type the command correctly.
- You see what normal output looks like.
- You understand how long different recovery jobs take.
- You can move your own seed into a safer environment only after you are confident with the tool.
This is especially useful when you are dealing with something as sensitive as a mnemonic phrase. A little practice now saves a lot of stress later.
How BTCRecover treats extra words
The key idea in this recovery method is that BTCRecover can treat an unknown extra word like a “big typo.” You tell the tool how many extra words you suspect are present, and it works through possibilities by removing those words from different positions in your best-guess mnemonic.
In the example from the video, the backup has 34 words instead of 24. That means there are ten suspected extra words. So the relevant settings are:
--big-typos 10--mnemonic-length 24
If you already know your seed should be 24 words, you may not always need to set the mnemonic length manually, because BTCRecover can assume a 24-word seed when it receives too many words. But if you are working with a shorter mnemonic, such as a 12-word seed, setting the expected mnemonic length becomes more important.
For example, if you have a 12-word seed plus ten extra words, your backup may contain 22 words total. If BTCRecover does not know that the real mnemonic should be 12 words, it might misinterpret the job and fail to find the result you are looking for. In cases like that, explicitly telling it the expected mnemonic length helps.
Common command-line options explained
The recovery command uses several options that matter a lot in practice. Here is what I pay attention to:
-
Wallet type
You tell BTCRecover that you are working with a BIP39 wallet. This avoids unnecessary prompts and makes the run cleaner. -
Derivation path
This is one of the most important settings. The derivation path tells BTCRecover what address format to use, what coin you are looking for, and which account to search in. If your target address comes from a specific wallet or exchange setup, this needs to match that setup. -
Address limit
This controls how many addresses are checked after the first one in the account. I often start with an address limit of 10 because many people find their earliest useful Bitcoin address within about ten positions from the first address in the account. If you need to search further, increasing the address limit can help, but it also increases work. In some cases, using an address database is a better approach than blindly raising the limit very high. -
Mnemonic length
This tells BTCRecover how long the real mnemonic should be. It matters most when you are not using a standard 24-word seed. -
Big typos
This is where you tell BTCRecover how many extra words it should try to remove. In the example, that number is 10.

Running the recovery for your target address
Once you have your best-guess mnemonic ready, you run the recovery against an address you know belongs to that wallet. The tool then tests possible corrected mnemonics by removing the suspected extra words and checking whether any of them produce addresses matching the one you are looking for.
In the video example, the command includes:
--wallet-type BIP39- a derivation path such as
m/49'/0'/0' --mnemonic-length 24--big-typos 10- the target address you are trying to match
The exact derivation path will depend on your wallet, coin, and account. Do not just copy a path from an example unless you know it matches your situation. If you have multiple accounts in Ledger Live or Trezor Suite, for instance, make sure you are looking in the correct one.

What to expect while it runs
Depending on how many extra words you need to remove, this can take anywhere from seconds to weeks. The tool will show progress as it counts possibilities, but the time estimate depends heavily on your machine and the size of the search space.
In the example with ten extra words, the run takes a meaningful amount of time. You do not need to sit and watch every second, but you should make sure the environment is stable enough that the process can complete without interruption.

The real limit: exponential complexity
This is the part I want people to understand most clearly.
Recovering extra words does not get slower by a small, linear amount each time you add another unknown word. It gets much worse very quickly because the search space grows exponentially. Each additional unknown word multiplies the number of possibilities BTCRecover has to consider.
In practical terms:
- A few extra words can be solved in seconds or minutes.
- Around ten extra words may take a long time, but it is still possible on suitable hardware.
- Twelve extra words can push into hours.
- Fourteen extra words can take days.
- Sixteen extra words can take weeks.
The exact times vary by machine and settings, but the shape of the problem does not change: more unknown extra words means dramatically longer recovery time.

This is why adding a handful of valid BIP39 words to your seed phrase is not a good security strategy. It may feel like you have made the backup harder, but from an attacker’s point of view it can still be very easy to brute-force if only a few extra words are involved.
A better alternative: use a BIP39 passphrase
If you want an additional layer of protection for your mnemonic, I would strongly recommend using a BIP39 passphrase instead of inserting random extra words into the seed phrase itself.
A BIP39 passphrase works differently from adding words to the mnemonic. It is not part of the 12 or 24-word backup you normally write down; it is an additional secret that changes the derived keys when combined with the mnemonic. That makes it a much cleaner and more practical way to add protection, as long as you understand how to back it up safely.
The important point is this: if your instinct is “I want my seed to be harder than just the written words,” do not solve that by making the mnemonic longer in an ad hoc way. Use a proper BIP39 passphrase and make sure you have a secure backup plan for both pieces of information.
A known error workaround: negative shift count
While making this video, I ran into some cases where BTCRecover produced an error ending with “negative shift count.” If you hit that error while following this guide, do not panic. It is a known issue in certain situations and may be fixed later as the Python 3 work continues.
For now, one workaround is to pad your best-guess mnemonic with one extra valid BIP39 word before running the recovery. That can make the job take longer, but it may allow you to continue if you are not already near the upper practical limits for this type of search.

If you run into problems while following these steps, check the video comments, send a message on Reddit, or open an issue on GitHub so I can help sort it out.
If you are stuck and need help
I would much rather you do this yourself in a safe environment than ask someone to handle your seed for you. But if you are completely stuck, frustrated, or worried about doing something wrong, there is structured information on my website about assisted calls and trusted recovery options: recovery services and consultations.
Please do not randomly email me your seed phrase out of the blue. Sending a mnemonic in plain text to someone you have just met online is terrible security practice, even if the recipient means well. If you need help with a real recovery, read through the website information first so you understand how pricing works and how to send sensitive material securely.
Conclusion
If your backup has more seed words than expected, do not assume it is hopeless. BTCRecover can often recover from this by treating unknown extra words as big typos and removing them until the mnemonic matches a known target address.
But the bigger lesson is that adding valid BIP39 words to your seed phrase does not add much practical security. A few extra words are very easy to brute-force, and each additional unknown word makes recovery exponentially harder for you if you ever need it. If you want stronger protection, use a BIP39 passphrase instead.
And whatever you do, practice with example seeds first, understand your derivation path, set the right mnemonic length, and only run the real recovery on a computer you trust.
Questions from viewers
Common questions from the comments on this video, with my answers.
What command-line arguments do I need if I am using a standard wallet and just want to recover the first account?
You shouldn't need to use any specific BIP39 paths if you are just using the first account and sticking with the defaults for any mainstream wallet. The tool will handle it automatically.
Can this tool recover a seed phrase where the words are out of order?
If the funds are out of order, you are likely stuffed. This tool works best when you have the correct number of words but some are missing or extra, not when the sequence is scrambled.
How should I configure the tool for an Ethereum address?
Yes, if the words are in order, the tool can help. You should specify that you have extra words rather than missing ones if that matches your situation, and use the appropriate wallet type for Ethereum.
What address format should I select for a Trezor Model One wallet with addresses starting with 'bc1q'?
Just select that you have a standard BIP39 wallet; it will automatically handle native segwit address types.
Will this work if I am missing only one word from my 12-word phrase?
Yep, one word will work with all the defaults and recover it in a few seconds, no problem at all.
How do I use this tool for a Cardano (Daedalus) wallet?
It should just work if you select the Cardano wallet type. Make sure you identify whether it is a Byron or Shelley era wallet, as that can affect the configuration.
Stuck with a wallet recovery?
I maintain BTCRecover and have helped people recover seed phrases, passphrases and wallet passwords since 2019. Book a one-hour video session and I’ll walk you through doing the recovery yourself, without sharing your seed. Or, if you’d rather I run it, a trusted recovery is charged only if it succeeds.
Private sessions: US$100 per hour. Trusted recovery: 10% of recovered funds. I never contact people first and will never ask for your seed phrase over chat or email.
Found this useful?
These guides are free and ad-free. You can support them by sharing them, subscribing on YouTube, or sending a tip.







