, , , ,

Brute Force BIP39 Passphrase Recovery (25th Word)

If you have the right seed phrase but cannot remember your BIP39 passphrase, this is a practical follow-up on how to approach recovery with BTCRecover. In the video I walk through several examples using short passphrases, dictionary lists and common password lists. You can also read this as a standalone guide: Watch the video.

The important thing to understand up front is that this is not magic. It is a structured guessing process. If your passphrase was short, simple or based on common words, recovery may be realistic. If it was long and truly random, brute force becomes much harder, often impractical without serious hardware. That said, running the right tests can still help you confirm what you have tried and narrow down possibilities.

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.

What You Need Before Starting

Before touching BTCRecover, make sure you have these prerequisites:

  1. A correct BIP39 seed phrase.
    This is usually 12 or 24 words. In this video I use a 24-word example. If the seed itself is wrong, no amount of passphrase testing will help.
  2. At least one known address derived from that wallet.
    BTCRecover needs something to test against. For Ethereum examples, that means an ETH address you know belongs to the wallet. For Bitcoin or other chains, it would be the equivalent known output/address.
  3. A working recovery environment.
    You should have BTCRecover installed and able to run tests. If security is a concern, use an isolated or air-gapped setup where possible.

If you are unsure whether your seed phrase is correct, stop there and verify it first. A wrong seed will waste hours of testing time.

Two Main Attack Approaches

There are two broad ways to approach forgotten BIP39 passphrase recovery:

1. Brute Force Token Attacks

This is where you give BTCRecover a file of individual characters, such as letters and numbers, and let it combine them in different orders. This works best when the forgotten passphrase is short.

For example, if someone used a simple four-character passphrase for plausible deniability or convenience, a token-based brute force attack can be very effective. The tool tries combinations from your token list until it finds one that matches the known address.

The downside is speed. As the possible length increases, the number of combinations grows quickly. A short test may finish in minutes. A longer one can take days or weeks depending on hardware and settings.

2. Dictionary and Password List Attacks

This approach uses lists of likely words or passwords instead of raw characters. You might use:

  • An English dictionary
  • A list of common first names or surnames
  • A leaked password list such as RockYou
  • A custom list made from family members, pets, nicknames or other personal terms
  • Diceware-style word lists if you used that method

Dictionary attacks are often better when the passphrase is longer but not fully random. For example, a phrase built from common English words may be far easier to find with a dictionary than by brute forcing every possible letter combination.

You can also combine dictionary words in a limited brute force style, such as trying two or three words strung together. This is powerful, but it hits practical limits quickly if the list is too large or you ask for too many words.

Hardware Speed Matters A Lot

Recovery time depends heavily on the hardware you use. In my tests I used a mix of dual-core and quad-core i5 systems, as well as a 48 vCPU Linode instance. The difference was significant.

A high-end consumer CPU such as a Ryzen Threadripper can be dramatically faster than an ordinary laptop or desktop processor. If a test would take months on slower hardware, it may become feasible on stronger equipment. That is why you should always look at the estimated time remaining and compare it against the machine actually running the job.

When judging whether a recovery attempt is realistic, ask:

  • How many cores are being used?
  • Is the process running flat out?
  • What is the ETA for the full search space?
  • Could a faster CPU or cloud server make the difference between months and days?

Do not assume one person’s timing will match yours exactly. Your hardware, list size, typo checks and token settings all change the result.

Example 1: Brute Forcing a Short Passphrase

In the first example, I tested a short passphrase using a simple token list containing lowercase letters and numbers. The goal was to recover a known Ethereum address from the same seed phrase.

The command used a token file with individual characters and limited the search to four tokens. Because the target was short, BTCRecover found it quickly. In this case, the test completed in about a minute, although the full run would have taken around an hour if the answer had been at the end of the list.

This is the kind of situation where brute force makes sense: the passphrase is short, the character set is limited, and you are not adding heavy typo checks.

Example 2: Capitalization Typo Checks

The second example used a similar token list but added capitalization handling. This matters because people often remember a word correctly but get the casing wrong.

For instance, a passphrase might contain a capitalized letter in the middle of a word, or the first letter may be uppercase when you expected lowercase. BTCRecover can test variations like that, but it increases runtime significantly.

In my test, adding capitalization typo checks changed the run time from about one minute to around five minutes for the same short search space. That is still manageable, but it shows why you should not enable every possible typo option by default.

If you are doing a broad search, be selective. If you have a strong idea of what the passphrase was and only need to account for small mistakes, targeted typo checks can be useful. If you are brute forcing from scratch, too many typo options can make the test explode in complexity.

Why Max Tokens Matters

One of the most important settings is max tokens. This tells BTCRecover how many items it may combine when using a token list.

For example:

  • If your token file contains single letters and you set max tokens to 4, it will try combinations up to four characters long.
  • If you are chaining dictionary words together and want two-word phrases, you would set max tokens to 2.
  • For three-word phrases, you would set it to 3.

This setting is critical. If you leave it too high, BTCRecover may start testing far more combinations than you intended. A test that should finish in hours can become one that takes days or weeks almost immediately.

Before starting a long run, double-check:

  1. What is the token list?
  2. How many tokens are you allowing?
  3. Are typo checks enabled?
  4. Is the known address correct?
  5. Have you tested the command on a smaller version of the same job?

A short test first can save you from wasting days on a misconfigured run.

Building a Token List

The token list used for the early examples was very simple: one character per line, covering lowercase letters and digits. The exact contents should depend on what passphrase style you think is likely.

You need to decide whether your forgotten passphrase might include:

  • Lowercase letters only
  • Uppercase letters
  • Numbers
  • Symbols
  • Spaces or other separators
  • Words from a specific language or list

The larger the token set, the faster the search space grows. If you are unsure, start with a smaller, plausible character set rather than including every possible symbol at once.

It is also worth checking what your hardware wallet actually allows. Different devices have different passphrase input restrictions. For example:

  • A Trezor may allow keyboard entry on your computer and support a wider range of symbols.
  • A Ledger requires on-device entry and offers a smaller selectable character set.
  • Other wallets such as Keepkey may have their own limits depending on firmware and interface.

If your device does not let you enter certain characters, there is no point including them in the token list. Build the search space around what was actually possible to enter in the first place.

Token list file with single characters shown next to example BTCRecover commands
A simple token list contains one character per line, while the side panel shows example commands.

Dictionary and Password List Examples

The later examples move from pure brute force into dictionary-style attacks. These are useful when the passphrase is likely to be made of real words or common password patterns.

In one test, I looked for a common surname using three different methods:

  1. A brute force token attack with capitalization handling
  2. An English dictionary list
  3. A leaked password list such as RockYou

The results showed why password lists can be so effective. The brute force approach took hours on the desktop CPU used in the test, while the dictionary and password list approaches found the target much faster. In one case, a password list located the answer in seconds.

This is not just about speed. It is also about realism. Many people choose passphrases that feel unique to them but are actually common choices. Leaked password lists contain millions of examples of what humans tend to pick when they think something is random. Checking against such lists is often worthwhile, even if you believe your passphrase was unusual.

You can also combine a password list with limited typo checks. For example, BTCRecover can test variations where certain letters are capitalized. In one demonstration, this helped find a longer familiar term that had non-obvious capitalization. The targeted run finished in minutes, though the full search space would have taken days if the answer were at the end.

The lesson is simple: password lists scale more predictably than token combinations. A larger list takes longer, but not exponentially in the same way as adding another word to a token-based phrase.

Chaining Two and Three Dictionary Words

Another powerful technique is chaining multiple dictionary words together. For example, you might test two-word phrases from a limited dictionary, or three-word phrases from an even shorter list.

In one example, I used a small numbered word list and searched for a two-word passphrase. The target was found quickly because the dictionary was constrained and no heavy typo checking was needed. This illustrates how useful short dictionaries can be when you have reason to believe the phrase came from a limited set of words.

In another example, I tested three-word phrases using a shorter English word list. On a laptop, that run took about two days. On faster hardware it would have been much quicker: roughly one day on a desktop and around three hours on a high-core-count system.

This is where limits become important. Chaining words sounds simple, but the math grows fast. If you use a large dictionary and ask for too many words, memory and computation requirements can quickly become unmanageable. In one test using a larger English token list with a long passphrase, the run simply ran out of memory and did not work.

Practical Limits: Memory and Computation

When testing BIP39 passphrases, you are constantly balancing three things:

  • List size
  • Number of tokens or words allowed
  • Typo variations enabled

Password lists scale roughly linearly. If a list has twice as many entries, the test takes about twice as long. If it is ten times longer, it takes about ten times longer. That makes password lists easier to estimate and plan around.

Token lists are different. They scale exponentially with both list size and token count. Doubling the token list can make the search much larger than double. Increasing the allowed number of words by one can multiply the work dramatically.

For example, if a dictionary has 100 possible words:

  • A two-word phrase has far fewer combinations than a three-word phrase
  • A four-word phrase has many more again
  • Each extra word multiplies the search space

This is why max tokens matters so much. It is not just a minor setting. It controls how large the test becomes.

Slide explaining computational limits for password lists and token lists
Password lists scale linearly, while token lists can scale exponentially as list size or word count increases.

Suggested Recovery Workflow

Here is the workflow I would recommend if you are trying to recover a forgotten BIP39 passphrase.

Step 1: Set Up and Test First

Before running a long recovery job, set up your environment properly and run some small test cases. This helps confirm that:

  • BTCRecover works
  • Your seed phrase is entered correctly
  • The known address is correct
  • Your token or password list files are valid
  • You understand the command arguments

This step can save you from running a multi-day job that fails because of a typo, missing file or misunderstood setting.

Step 2: Start With Short Brute Force If Appropriate

If you have no strong idea what the passphrase was, start with a short brute force search using a limited token list. This is useful if the original passphrase might have been simple, memorable or chosen mainly for plausible deniability.

Use a small character set and keep max tokens low at first. If something short works, you may be done quickly.

Step 3: Try Dictionary and Password Lists

If brute force does not work, move to dictionary-style attacks. Try several different lists:

  • Common English words
  • Names and personal terms
  • Leaked password lists
  • Diceware or other word-based lists if relevant

You can run these as straightforward password list tests first. If the passphrase was a single common entry, this may find it quickly.

Step 4: Build a Personal Token List

If you suspect the passphrase was made from personal words, create your own token list. Include things such as:

  • Family member names
  • Nicknames
  • Pet names
  • Friend names
  • Common password words you might have used
  • Short phrases or memorable terms

Then run limited multi-token tests against that list. Many people build passphrases by combining two or three familiar items, so this can be a realistic approach.

Step 5: Be Selective With Typo Checks

Do not enable every typo option unless you have a reason to. Missing letters, doubled letters, caps lock mistakes and other variations are useful when you already have a strong candidate in mind. But if you are doing broad brute force or dictionary testing, too many typo checks can make the run extremely slow.

Use typo checks strategically:

  • First test the base list without heavy typo options
  • If that fails, add targeted capitalization handling
  • Only then consider broader typo variations if your hardware and time budget allow it

Cloud Servers and Security Risks

If a recovery job is likely to take months on local hardware, using a faster cloud server or high-end CPU can be cost effective. In my tests, moving from an ordinary desktop to a much larger system reduced the estimated run time dramatically.

However, you must understand the risk. Running BTCRecover exposes both your seed phrase and the candidate passphrases to the machine doing the work. If that machine is rented cloud infrastructure, you should treat it as untrusted after the job finishes.

If you use a cloud server for recovery:

  1. Assume the seed phrase may have been exposed
  2. Assume the recovered passphrase may also be compromised
  3. Move funds to a brand-new wallet afterward
  4. Use a new 24-word seed and a new passphrase

Do not simply move funds into another address from the same compromised wallet. The safest path is a completely fresh setup where none of the old secrets remain in use.

Slide about cloud server risks and suggested workflow
Using a cloud server can speed up recovery, but you should assume exposed secrets may be compromised.

Is Brute Force Passphrase Recovery Worth It?

Brute force BIP39 passphrase recovery is a low-odds play. There are no guarantees of success, especially if the forgotten passphrase was long and random. If that is your situation, manual memory work, checking old notes, email history or device backups may be more useful than spending months on computation.

That said, running structured tests can still be worthwhile. It helps you confirm what has been tried, avoid repeating mistakes, and possibly recover funds if the passphrase was simpler than you remember. For many people, a short brute force test followed by dictionary and password list checks is a reasonable first pass before giving up or seeking outside help.

If you are dealing with lost funds and need guidance on recovery options, my wallet recovery services page covers when professional help may make sense.

Final Takeaways

  • You need a correct seed phrase and at least one known address to test candidates
  • Short passphrases can sometimes be recovered with brute force token lists
  • Longer or word-based passphrases often respond better to dictionary and password list attacks
  • max tokens controls how many characters or words BTCRecover will combine, so set it carefully
  • Password lists scale roughly linearly, while token combinations scale exponentially
  • Capitalization and typo checks can help, but they increase runtime significantly
  • Check your hardware wallet’s passphrase character limits before building a token list
  • If you use a cloud server, move funds to a completely new wallet afterward

The goal is not just to run one big command and hope. It is to build a sensible sequence of tests that matches what you actually remember, what your device allowed, and what kind of hardware you can realistically use.

Questions from viewers

Common questions from the comments on this video, with my answers.

Does BTCRecover need an internet connection to check if a guessed passphrase is correct?

No, the software does not have any networking capabilities and does not rely on checking balances online. It works entirely offline by deriving addresses from your seed and passphrase locally. You can either provide specific known addresses for it to match against, or run it against a local Address Database to see if the generated address has ever received funds on the network.

Is it safe to run the recovery process on a computer connected to the internet?

You should not run BTCRecover on an internet-connected PC when using your actual seed. The best practice is to install the software and get familiar with how it works first, then disconnect all networking before running it with your real seed phrase. Ideally, you should keep the machine offline until after you have moved your funds to a new seed.

What are my options if I have forgotten my passphrase completely and don't know what it might be?

If you have no idea what the passphrase could be, your only option is to attempt a recovery based on passwords you may have used elsewhere, a dictionary attack with common English words and phrases, or pure brute-force for very short passphrases (such as 5 letters or less, or 10 digits or less). If the passphrase was long or random, it likely cannot be recovered without a backup. If you’d like hands-on help, see wallet recovery services.

How do I handle special characters like '$' in my token list file?

You need to read the documentation on how to include special characters in your tokenlist. For example, if you want to search for a dollar sign between two words, you must format the file correctly according to the syntax rules for expanding wildcards and special characters found in the official docs.

Can I reduce the number of BIP39 words I have to check by removing some from the word list?

Yes, you can use a tokenlist approach similar to a scrambled seed. Instead of editing the main language file directly (which causes errors), you create a custom tokenlist file containing only the specific words you think might be in your seed. You then specify the language in the arguments and use positional and required anchors for the positions where you know the exact words.

Which hardware is better for running these recovery scripts: a MacBook Pro M2 or a standard laptop with an i5 processor?

The MacBook Pro M2 will be faster. The performance on these Apple Silicon devices is excellent for this type of processing compared to a regular laptop that lacks a dedicated GPU.

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.

Stuck on this?

Book a one-hour session and I’ll walk you through it, or ask about a trusted recovery.

Recovery guides

More guides