, , ,

Picking a Good BIP39 Passphrase (25th Word) or Avoiding a Bad One

Over the last few weeks, I’ve been looking at how to recover a BIP39 passphrase using tools like BTCRecover. That work coincided with some news and drama around a key-extraction attack that affected Trezor wallets, and it caused a lot of confusion on Reddit.

In this article, I’ll walk through the assumptions you should avoid when choosing a BIP39 passphrase, also known as the 25th word. Then I’ll give practical suggestions for creating something that is secure enough, but still realistic to remember and back up over many years.

Watch the video

Assumptions to avoid when choosing a BIP39 passphrase
A slide titled ‘Assumptions to Avoid’ appears next to a YouTube playlist of crypto recovery tools.

Assumptions to avoid with BIP39 passphrases

The biggest mistake people make is assuming that a BIP39 passphrase behaves like an online password. It doesn’t. A BIP39 passphrase protects access to your seed, and if someone has both your 24-word backup and your passphrase, they can recover your funds. That means the security model is very different from logging into email, Facebook, or online banking.

Here are some assumptions I’ve seen repeatedly that you should not rely on.

“Cracking tools are hard to find”

They are not. Tools for recovering a BIP39 passphrase are easy to obtain and open source. You can run them on Windows with minimal skill. Also, an adversary may have access to proprietary tools that are faster or more efficient than the open-source options I demonstrate in my videos.

Do not assume that because you cannot easily find a tool, someone else cannot.

“I use Ledger, so key extraction is not my problem”

If you have a Ledger wallet, you should not assume that key extraction is impossible just because we do not know about vulnerabilities for Ledger wallets today. The important point is that a passphrase can still protect your 24-word seed backup if it falls into the wrong hands.

A hardware wallet protects your keys while in use. A passphrase adds protection to the recovery material itself. Those are two different layers.

“My online password is short, so my BIP39 passphrase can be short too”

There is a huge difference between an online account and a BIP39 passphrase.

With an online service, the provider may detect brute-force attempts, lock accounts, require verification, or otherwise respond to suspicious activity. With a BIP39 passphrase, someone can attack it offline without you knowing. There is no login page warning you that someone is trying thousands of guesses per second.

That makes short passphrases much more dangerous than they appear.

Offline password attacks are different from online password attacks
A comparison between offline and online password attacks, with common passphrase misconceptions listed.

“My memory is reliable”

This one hurts because I see it over and over. People say, “I memorized my passphrase. It’s long enough. I’ll be fine.”

Then they search Reddit for “forgot passphrase,” and the results are not encouraging.

Reddit posts about forgotten BIP39 passphrases
A Reddit search for ‘forgot passphrase’ shows multiple users who lost access because of a forgotten BIP39 passphrase.

You may remember your passphrase today. But if you only enter it occasionally, your memory in two, three, or five years may not be as reliable as you think. A BIP39 passphrase is not just a security decision; it is also a long-term backup decision.

“If it is secure now, it will be secure for 10 years”

Computing power continues to increase. Moore’s Law has slowed in some areas, but we should still consider future computing capability when protecting funds we may hold for many years. A passphrase that feels strong today may not feel as strong a decade from now.

“Everyone needs a very long fully random passphrase”

Some people interpreted comments around the Trezor key-extraction attack to mean everyone needs a 37-character fully randomized passphrase before they are secure. That is not what I am saying.

The purpose of a BIP39 passphrase is not to recreate the full security of a 24-word seed phrase. It is an additional layer on top of your seed phrase. The right length depends on your threat model, your backup method, and how much time you want the passphrase to buy you if someone gets access to your 24 words.

“Everyone needs a passphrase”

Not everyone does. If you are not storing a large amount of cryptocurrency, and you can physically secure both your hardware wallet and your 24-word backup in a way you trust, you may not need a BIP39 passphrase.

Once you start using passphrases, you move into advanced territory. That is why some wallets disable this feature by default. It is easy to make a mistake and lose access to funds. The goal is to understand your risks and manage them based on your own circumstances, not copy someone else’s setup blindly.

Ledger discussion of passphrase length and mitigation
A Ledger page discussing mitigation and passphrase length appears next to a list of assumptions.

Who actually needs a BIP39 passphrase?

Before choosing one, ask yourself what you are trying to achieve. A passphrase is not just “extra security.” It has specific use cases.

You want protection if your 24-word backup is stolen

If someone gets access to your written seed, a passphrase can stop them from using it immediately. This is especially important if your backup is stored somewhere that could be physically compromised.

Ask yourself: how quickly would you notice if your 24-word backup or hardware wallet was stolen? If the answer is “I don’t know” or “probably not for weeks,” a passphrase gives you more time to react.

You store part of your recovery material with someone else

If you keep a copy of your seed with a friend, family member, lawyer, or in another location, a passphrase can reduce the risk if that person’s storage is compromised. They may have one piece, but they do not have both pieces needed to access funds.

You want plausible deniability

Plausible deniability means you can unlock a wallet and show some balances without revealing a second hidden wallet. The visible wallet might be accessible with just the 24 words, while a larger or more important wallet requires the passphrase.

This is useful in situations where someone could physically force you to reveal access to a device. It does not solve every problem, but it can add an extra layer of privacy and safety.

You want backup protection against future hardware attacks

Even if your current wallet model has no known key-extraction issue, that may change. A passphrase protects the recovery material itself, which is useful if future vulnerabilities are discovered or if proprietary attack tools become more effective.

Defining your passphrase objective

The most important question is: how much time do I want my BIP39 passphrase to buy me?

You need to decide what level of risk you are comfortable with. For example, you may be asking:

  • How long would it take an attacker to have a 1% chance of finding my passphrase?
  • Are they using a CPU, GPU, or hypothetical ASIC?
  • Do I care about protection for one month, one year, five years, or longer?
  • Is this just for plausible deniability, or is it protecting a large backup?

Brute-forcing passphrases is an exercise in probability. You are not trying to know the exact time it will take in every situation. You are choosing a level of risk you can live with.

Questions to ask when selecting a BIP39 passphrase
A slide lists questions and points to consider when selecting a BIP39 passphrase.

The most important rule: choose something you will remember and back up

The second most important thing, after defining your objective, is picking a passphrase you will not forget. In practice, this may be the most important thing of all.

A BIP39 passphrase does not have the same typo resistance as a 24-word seed. A standard BIP39 wordlist is designed to make backups more forgiving. If you write down “aband…” instead of “abandon,” it is usually obvious what was meant, and many tools can recover from minor errors if enough context exists.

A passphrase with random characters, symbols, capitalization, spaces, or unusual words does not have that same safety net. One small transcription error can mean you lose everything.

That means your backup method matters as much as the passphrase itself.

Backup options to consider

You might use one of these approaches:

  1. Write the 24-word seed on paper and store it securely.
  2. Store the passphrase separately, such as in a password manager or another electronic location.
  3. Use durable metal backup storage for both the seed and passphrase if you want stronger physical protection.
  4. Keep separate backups of the two pieces so an attacker would need to compromise multiple locations.

The idea is that someone needs both your 24-word seed and your passphrase. If one piece is stolen, they should not have access to funds without the other.

ColdTi titanium backup storage
An Amazon product page for a ColdTi titanium cryptocurrency seed storage wallet.
CryptoSteel MNEMONIC backup storage
An Amazon product page for the CryptoSteel MNEMONIC cold storage wallet.

Avoid reused, short, or leaked passphrases

Do not reuse passwords from online services. The last thing you want is to discover that an old account was hacked and your BIP39 passphrase has been linked to your name or email address in a leak database.

Also avoid very short passphrases. In my previous work on brute-forcing, I showed how easy it can be to attack something that is six characters or less. That may sound acceptable for some low-value online accounts, but it is not acceptable for protecting cryptocurrency recovery material.

If you invent a passphrase from your imagination, check known leak lists such as RockYou.txt and phpbb.txt. A phrase you think is original may have been used by many people before. If it appears in a dictionary or leak list, it becomes much easier to break with a dictionary attack, even if the phrase looks long or complex.

Checking a passphrase against the RockYou leak list
A text editor shows entries from the rockyou.txt password leak list.

My practical recommendation: use lowercase Diceware words

For most people, I recommend using only Diceware-style words and keeping everything simple. That means no uppercase letters, no symbols, no spaces, and no clever punctuation.

Why? Because complexity often creates more backup problems than it solves security problems. In my experience, you are usually better off adding another word than making the passphrase harder to write down correctly years later.

Why lowercase words work well

Lowercase Diceware passphrases have several advantages:

  • They are easier to back up accurately.
  • They reduce transcription mistakes.
  • They avoid confusion between similar characters, such as capital I and lowercase l.
  • They make it easier to store in a password manager or physical backup.
  • You can increase security by adding more words instead of making the format fragile.

You should also try to balance randomness. Do not just use song lyrics, book sentences, or predictable phrases. The goal is something that is memorable enough for you, but not so predictable that it appears in common dictionaries or leak lists.

BIP39 wordlist shown while discussing passphrase choices
A text editor displays the beginning of a numbered BIP39 wordlist.

Using EFF Diceware lists

If you use Diceware, I suggest looking at the newer EFF wordlists because they include improvements over older lists. It is also useful to understand the trade-offs between the short and long lists.

The long list gives you more security per word because it has a larger dictionary size. The 2.0 short list can make backups easier to manage, which matters when reliability is your biggest risk.

You need to decide what you are optimizing for: maximum raw entropy, or a backup method that will survive human error over many years.

EFF article on new wordlists for random passphrases
The EFF article about new wordlists for random passphrases is shown.

Spreadsheet model and security recommendations

I put together a spreadsheet to estimate how long it would take different hardware to reach a 1% chance of finding a passphrase made from Diceware-style words. The model looks at dictionary size, number of words, total combinations, and estimated attack time on:

  • A 48-core CPU server
  • A GPU
  • A hypothetical ASIC

The most useful columns are the ones showing how long it would take to reach a 1% chance of success. That is because brute-forcing is probabilistic. You are not asking, “Will this exact attack succeed?” You are asking, “What level of risk am I comfortable with over time?”

Spreadsheet estimating BIP39 passphrase brute-force times
A spreadsheet shows passphrase brute-force time estimates.

The assumptions include using performance data from my previous BTCRecover work, assuming a GPU could perform at least 10 times faster than that baseline, and assuming an ASIC could perform about 1,000 times faster. I also assume that if you could recover something within a year using BTCRecover, someone with proprietary tools might do it in about a month.

There is also a useful article from the Sia team about their experience building an ASIC for crypto mining. It helped explain the economics behind ASIC development: if attacking BIP39 wallets becomes profitable enough, someone may build specialized hardware for it. That does not mean it will happen tomorrow, but it is worth considering when protecting funds for many years.

Sia article on ASIC mining economics
A Sia blog article about ASIC mining economics is shown with highlighted text.

Suggested passphrase lengths by goal

These are practical recommendations based on the spreadsheet model and a 1% chance threshold. They are not universal rules, but they give you a starting point.

For plausible deniability

If your main goal is plausible deniability, and you want an attacker to have only about a 1% chance of finding the passphrase within a month using a hypothetical ASIC, these options may be reasonable:

  • Five words from the new EFF short list
  • Four words from the long list

This level may suit people who mainly want a visible wallet plus a hidden wallet. It is not necessarily enough to protect a large backup against a determined multi-year attack.

For protecting your backup over several years

If you want stronger protection for your 24-word backup, and you are thinking about an attacker having only about a 1% chance of finding the passphrase within five years using a hypothetical ASIC, consider:

  • Six words from the EFF short list
  • Five words from the long list

Every additional word significantly increases security. If you can realistically back up five words, ask yourself whether you can manage six. Especially if your passphrase is written down or stored in durable backup media, adding one more word may be worth it.

Spreadsheet comparing dictionary sizes and word counts
A spreadsheet compares passphrase brute-force times across dictionary sizes and word counts.
Passphrase strength table showing 1% chance estimates
A table compares passphrase strength across dictionary sizes, word counts, CPU, GPU, and ASIC scenarios at a 1% chance.

How to choose your BIP39 passphrase step by step

Here is a practical process you can follow.

  1. Decide why you need a passphrase.
    Are you protecting a backup? Adding plausible deniability? Preparing for future hardware vulnerabilities? Your answer changes the length and strength you should target.

  2. Choose your acceptable risk level.
    Decide what probability of discovery you are comfortable with, and over what timeframe. For example: 1% chance in one month, one year, or five years.

  3. Pick a word list.
    If using Diceware-style words, consider the EFF short or long lists. Understand that longer lists provide more entropy per word, while shorter lists can make backups easier to manage.

  4. Use lowercase words only.
    Avoid uppercase letters, symbols, and spaces unless you have a very strong reason. Simplicity reduces backup errors.

  5. Avoid reused passwords and common phrases.
    Do not reuse online-service passwords. If you invent your own phrase, check it against known leak lists such as RockYou.txt and phpbb.txt.

  6. Make the passphrase long enough for your threat model.
    For plausible deniability, five EFF short-list words or four long-list words may be a starting point. For stronger backup protection over several years, six short-list words or five long-list words are better.

  7. Back it up in a way that survives mistakes.
    Your passphrase needs to survive transcription errors, physical damage, and your own future memory gaps. Consider separate storage for the seed and passphrase, durable metal backups, and a password manager where appropriate.

  8. Test recovery before you rely on it.
    Make sure you can restore access using both your 24-word seed and your passphrase. If you cannot test this safely in a way that does not expose funds, at minimum verify that your backup materials are complete and legible.

  9. Review annually.
    Security assumptions change. Hardware changes. Your storage situation may change. Revisit your passphrase and backups regularly.

Final thoughts

The key is to balance security with reliability. Do not choose something so complicated that you will be back on Reddit in 12 months asking how to recover from a mistake you made while backing it up.

A good BIP39 passphrase should do three things:

  • It should match your actual threat model.
  • It should be easy enough to back up accurately.
  • It should remain usable and memorable over many years.

If you are already dealing with a lost or forgotten BIP39 passphrase, this is the point where the problem becomes much harder than choosing one. In that case, see recovery services and consultations for help understanding your options.

Questions from viewers

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

Is a BIP39 passphrase the same thing as my seed phrase or mnemonic?

No, they are different. The seed (or mnemonic) is your primary backup of the wallet's keys. The BIP39 passphrase is an additional layer that, when combined with the seed, generates a completely different set of keys. You can think of it as a '25th word' that creates a hidden wallet separate from your main one. If you lose the passphrase, you cannot access the funds associated with that specific hidden wallet, even if you have the seed.

How do I verify that my passphrase is working correctly without losing my funds?

You should test it before sending significant amounts of crypto. On devices like Ledger, you can use a 'temporary' passphrase mode to check if it works. Unlock your device with your normal PIN, assign the passphrase in temporary mode, and then try to receive funds to an existing account. If the address matches what you expect on the device, it is working. Some wallets also allow you to re-assign the passphrase to a PIN without wiping the device, which overwrites the old setting.

Should I use special characters or spaces in my passphrase to make it stronger?

I generally advise against using special characters because they make the passphrase more error-prone and harder to recover from transcription errors. It is better to rely on entropy by adding more Diceware words rather than 'spicing up' a weak password with symbols. Regarding spaces, while they are valid characters, I suggest avoiding them to keep things simple and reduce the chance of input errors. Just concatenate the words or use a clear separator if necessary, but prioritize length and randomness over complexity.

If my hardware wallet is stolen, does the thief need my passphrase to restore the wallet?

Yes, if you have enabled a BIP39 passphrase, the thief will need both the seed and the correct passphrase to access the funds associated with that hidden wallet. However, this is not a major security concern for most people because all decent hardware wallets support passphrases. If you only have the seed and no passphrase, they can access the main wallet. The passphrase protects the *hidden* wallet derived from that combination.

What is the maximum length for a BIP39 passphrase?

There is no universal standard maximum length; it depends on the specific hardware wallet vendor. For example, Trezor limits passphrases to 50 characters, while Ledger and Coldcard allow up to 100 characters. You should check the specifications for your specific device to ensure your chosen passphrase fits within their limits. (Apps and tools change, so check the current version if something looks different.)

Can I store my passphrase digitally if my seed is stored physically?

Yes, you can store the passphrase digitally, provided your seed backups remain non-digital (physical). This allows you to distribute seed backups to trusted individuals or locations without fully trusting them, as they would still need the digital passphrase to access the hidden wallet. However, I still recommend having a physical backup of the passphrase somewhere safe, just not stored with any of the seed backups, in case you lose access to the digital service.

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