Seeds, passphrases and the settings that decide whether you keep your coins
This page is for anyone about to create their first Sparrow wallet, or anyone who has one and
is unsure what the passphrase field, the word count and the gap limit actually control. The
short version: the words are the money, the passphrase changes which wallet those words open,
the wallet password only protects a file, and a restored wallet showing the wrong balance is
almost always a scanning setting rather than a theft.
Every step here exists because someone lost money by skipping it. Step six is the one
almost nobody does, and it is the only step that proves the other five worked.
Work offline if you can, on a machine you trust, with the door shut and no camera in the
room.
Create the wallet file and give it a name you will recognise later
Sparrow keeps each wallet as its own file. Name it for its job — savings, spending, testnet — because in two years you will have several and the file name is the only label visible before you unlock one.
Choose the script type before you write anything down
Native segwit is the sensible default for a new single-signature wallet: lowest fees, widest acceptance. The choice is recorded in the wallet’s output descriptor, the compact text formula that says which keys and which address type make up this wallet. The same seed with a different script type produces different addresses, so note the descriptor as part of your backup.
Generate the BIP39 mnemonic and copy it onto paper by hand
Sparrow generates a BIP39 mnemonic — the numbered list of English words that is your wallet. Write it with a pen. No photographs, no password manager notes typed on the same machine, no cloud documents, no screenshots. Copy the words in order, then read them back against the screen once more before closing the dialog.
Decide about a passphrase now, not later
A BIP39 passphrase is optional and it is not a second-chance safety net — it changes which wallet the seed produces. If you want one, add it at creation and record it separately from the words. If you are not certain you can store an extra secret reliably for a decade, leave it empty. Details in the passphrase section below.
Set the wallet password
The password encrypts the wallet file on disk. Sparrow uses Argon2, the 2015 Password Hashing Competition winner, tuned so that each attempt takes at least half a second on modern hardware, which makes brute-forcing a decent password expensive. Use something long and unique. This password is not a backup of your coins.
Restore-test the backup while the balance is still zero
This is the step people skip and the step that saves them. Close the wallet, create a second wallet from the words you wrote down (plus the passphrase, if you used one), and check that the first receive address and the extended public key match the original exactly. If they do not match, your backup is wrong and you have found out for free.
Only now send a small amount, confirm it, then fund properly
Send a token amount first, watch it confirm, then spend it back out to prove you can sign as well as receive. Practise the whole cycle on testnet first if you want a rehearsal that costs nothing — see the network section below. After that, pair a hardware signer if the balance justifies it.
12 or 24 words: the answer is not the one you expect
Both are BIP39 mnemonics and Sparrow treats them identically. The difference is
entropy: a twelve word phrase encodes 128 bits of randomness, a twenty-four word phrase
encodes 256. That sounds like a doubling of security, and in a narrow mathematical
sense it is.
It does not matter in practice. 128 bits is not a number that gets brute-forced by
anyone, with any budget, using any hardware that exists or is planned. Doubling an
already unreachable figure does not change your risk profile. What does change is the
amount of information you must copy without error, store without loss, and read back
correctly under stress — possibly by someone else, years from now, who has never seen a
seed phrase before.
Every self-custody loss worth studying comes from that side of the ledger: a
transposed word, a smudged sixteenth line, one plate stamped in the wrong order, a
backup that was never tested. Twenty-four words gives you twice as many chances to make
that mistake.
The recommendation: take twelve words for a wallet you will handle
yourself, and spend the effort you saved on a durable medium and a real restore test.
Take twenty-four if a hardware device generates them for you and you never plan to
transcribe them by hand more than once, or if you are matching a policy that other
signers in a multisig quorum already use. Either way, verify.
12 words
128 bits of entropy. Standard BIP39. Faster to write, verify and stamp.
24 words
256 bits of entropy. Standard BIP39. Twice the transcription surface.
Attack difference
Negligible — both are far past any feasible search.
Human error difference
Large, and it favours the shorter phrase.
What actually protects you
A tested backup on a medium that survives water, fire and moving house.
The BIP39 passphrase: the most useful and most dangerous field in Sparrow
Sparrow supports an optional BIP39 passphrase, the thing forums call the "13th word" or
"25th word". Almost everyone who loses funds to it does so because of one
misunderstanding, so start here: it is not a password on your wallet file.
What it actually does
Your mnemonic words are turned into the master key through a derivation function that takes
two inputs: the words, and the passphrase. Change the passphrase and you change the master
key, which changes every address, every key and the entire wallet. The words plus no
passphrase open one wallet. The same words plus cactus open a second, entirely
separate wallet. The same words plus Cactus open a third. There is no list of
valid passphrases anywhere, so nothing can tell you that you got it wrong.
The failure mode, stated plainly
If you forget the passphrase, a perfect twenty-four word backup is worthless. Not
difficult, not slow, not expensive — worthless. The passphrase is not stored in the seed,
not stored on your hardware device, and not recoverable by the Sparrow project, which has
no access to anything of yours. A single mistyped character produces a valid, empty wallet
that will happily generate addresses for you forever.
A wrong passphrase never produces an error
Enter the wrong passphrase and Sparrow shows you a working wallet with a zero balance and
no warning, because from the software's point of view nothing is wrong: that really is
the wallet those inputs derive. Before you trust a passphrase, restore from your written
backup on a clean wallet and confirm the first receive address matches character for
character. If the balance is zero and the addresses differ, the passphrase is wrong, not
the coins missing.
When it genuinely earns its risk
Two cases. First, if someone finds your paper seed, the words alone give them nothing —
this is the real argument for a passphrase, and it is a good one for anyone storing a
backup somewhere they do not fully control. Second, plausible deniability: the seed with no
passphrase can hold a small, real balance as a decoy while the passphrase-protected wallet
holds the savings. Under coercion you can hand over a working wallet.
Both benefits depend entirely on your ability to preserve an extra secret. If you are not
the kind of person who will still know exactly where that secret lives in eight years,
leave the field empty and protect the paper instead. An honest empty passphrase beats a
heroic forgotten one.
How to back it up
Separately from the words, or the passphrase adds nothing. If both live in the same
envelope, whoever opens the envelope has everything, and you have accepted the loss risk of
a second secret in exchange for no protection at all. Practical arrangements: words in a
home safe and passphrase with a solicitor; words on metal at home and passphrase memorised
plus sealed with a family member; words split between two properties and the passphrase in
a bank box. Write down where the other half is, in plain language, for whoever inherits
this.
Record it exactly: case, spaces, punctuation, and whether that trailing space you typed was
real. Note that a passphrase is a per-wallet choice — if you use one, the
multisig quorum or hardware signer you attach
it to must be given the same passphrase every single time you connect it.
Three different secrets
Password, seed, passphrase — people conflate all three
They protect different things, fail in different ways, and only one of them is your money.
Getting this straight prevents most support-forum tragedies.
The wallet password
Encrypts one wallet file on one computer. Sparrow uses Argon2 — winner of the 2015
Password Hashing Competition — tuned to require at least 500 milliseconds per attempt on
modern hardware, which makes offline guessing slow and costly.
If you lose it: that file is unreadable, and you recreate the wallet from
the seed. Inconvenient, not fatal. It is not a backup of anything.
The seed phrase
The 12 or 24 BIP39 words. This is the wallet. Anyone holding them can reconstruct your
keys in any compatible software, on any computer, forever — no password needed.
If you lose it: and the wallet file is also gone, the coins are
unrecoverable. This is the one thing you protect above all else.
The BIP39 passphrase
An optional extra input that selects which wallet the words derive. Not stored
anywhere, not verifiable, not recoverable. Its whole value is that it exists nowhere but
in your possession.
If you lose it: the funds in the passphrase wallet are gone even with a
flawless seed backup. Treat it with the same care as the words.
SLIP-39 is a different mnemonic standard that splits a master secret into several share
phrases using Shamir secret sharing, so that any threshold of them — three of five, say —
reconstructs the secret while fewer reveal nothing. It is attractive on paper: geographic
redundancy without any single sheet being sufficient.
Sparrow's documented seed support is BIP39 mnemonics, with Codex32 support added in the
2.4.0 release. SLIP-39 share sets are not part of that documented support, so do not build
a recovery plan that assumes Sparrow will reassemble them. If you already hold SLIP-39
shares from a device that created them, keep that device's own recovery path intact and do
not treat Sparrow as your fallback.
For the goal behind the question — no single location holding enough to steal, no single
loss being fatal — the tool Sparrow implements properly is
multisig. A 2-of-3 wallet gives you the same
threshold property at the level of the wallet policy rather than the level of a share
format, it is described by an output descriptor you can print, and any compatible wallet
can rebuild it. The trade-off is a more complex backup: you keep the descriptor as well as
the seeds.
Importing versus sweeping a private key
Sooner or later you will hold a loose private key: a paper wallet from years ago, a key
from a faucet or a gift card, a single address from an old app. There are two ways to
bring that money into your Sparrow wallet, and only one of them keeps your backup
honest.
Sweeping — the one to choose
Sweeping means spending everything that key controls into an address your own seed
derives. It is one ordinary on-chain transaction. Afterwards the coins live in your
wallet, covered by the seed backup you already tested, and the old key is irrelevant
even if a copy of it exists in someone else's drawer. You pay a fee once and your
recovery story stays a single sentence long.
In practice you need software that can sign with that key — the wallet that created it,
or a node you can import the key into — and you send the funds to a receive address
from your Sparrow wallet. Assume the old key is compromised until proven otherwise:
anything that has been typed into a website, printed by a service, or stored on a phone
you no longer control should be swept immediately rather than kept.
Importing — what it silently costs you
Importing keeps the key alive as a spending source that your mnemonic does not derive.
The balance shows up, everything looks fine, and your seed backup is now incomplete:
restore from the words and that money is missing, with no error message to explain why.
People discover this years later, which is the worst possible moment. Sparrow's software
keystores are built around BIP39 mnemonics, master private keys and watch-only extended
public keys precisely because a wallet should be describable by one backup.
Exporting and backing up keys and xpubs
What you back up is the mnemonic — viewable again in the wallet's own settings once you
enter the wallet password — plus the output descriptor that records the script type and
derivation. For watch-only use, Sparrow exports the extended public key, and it can
read and write Electrum wallet files if you are migrating.
Treat an xpub as sensitive, not public. It cannot spend, but it generates every address
your wallet will ever use, which means anyone holding it can see your entire balance
history, your incoming payments and your change — permanently, with no way to revoke
it. Do not paste it into a random block explorer, a portfolio tracker or a support
chat. If you want a watch-only wallet on another machine, move the xpub the same way
you would move a password.
The gap limit: the most common cause of "my bitcoin disappeared"
A wallet derived from your seed contains an effectively unlimited run of addresses. Nothing
can scan an infinite list, so wallets use a stopping rule: keep checking addresses in order
and give up after a set number of consecutive unused ones. That number is the gap limit,
and in Sparrow it defaults to 20 for normal wallets and 40 for postmix wallets.
Twenty is comfortable for ordinary use. It stops being comfortable when addresses got
consumed without receiving anything — you generated a batch of invoices that nobody paid,
an app churned through receive addresses, or you restored a wallet whose activity was
spread out over years and several devices. If more than twenty unused addresses sit between
your last known payment and the coins further along, the wallet stops before it reaches
them.
The coins are not gone. They are on-chain, controlled by keys your seed derives, visible to
anyone who looks at the right address. Your wallet simply never looked.
When a restored wallet shows zero or a partial balance
Confirm your server is connected and fully synced before concluding anything. A
catching-up node reports partial history that looks identical to missing money.
Check you restored with the same script type and the same passphrase state as the
original — a different derivation is a different wallet, not a scanning problem.
Raise the gap limit in the wallet settings — try 100, then several hundred if needed —
and rescan. Larger limits mean more queries and a slower refresh, which is the only
cost.
If the balance appears, leave the higher limit in place for that wallet so future
rescans behave.
Two habits prevent the whole problem. Do not generate receive addresses you are not
actually going to use, and label every incoming payment in Sparrow so that a future you can
reconcile the wallet against reality instead of guessing.
Sparrow supports mainnet, testnet, signet and regtest, and you switch with
Tools > Restart in Testnet. Testnet coins have no value and are free
from public faucets, which makes the whole first-time experience — receiving, coin
selection, fee choice, signing, watching a transaction confirm, bumping a fee — a
rehearsal that can cost you nothing.
Do it before your first real send. Ten minutes on testnet teaches you more about
Sparrow's transaction editor than any guide, and it moves your first genuine mistake to
a network where mistakes are free. Signet is the better choice if you want reliable
block times; regtest is for developers running a private chain.
Where Sparrow keeps its configuration
Sparrow stores its own settings — server connection, preferences, appearance, unit
display — in a configuration file inside its application data directory, separate from
your wallet files, which are individually encrypted. Wallets are kept per network, so
a testnet session does not touch your mainnet wallets. The exact directory differs by
operating system and by how you installed the application, so treat the official
documentation as the authority rather than a path copied from a forum post.
Two practical consequences. Backing up the configuration file is not backing up your
money; only the seed does that. And when you move to a new machine, copying wallet
files across is convenient but the seed remains the thing that must survive
independently of any computer.
Never mix networks or reuse a real seed for practice
A testnet address and a mainnet address are different animals, and sending real
bitcoin to an address from the wrong network is not a recoverable error. Read the
network indicator in Sparrow's window before you send anything, every time.
Equally: generate a throwaway seed for testnet practice. Do not type the mnemonic that
guards your savings into a wallet you are deliberately experimenting with, and never
type it into anything you did not
verify before installing.
The recommendation, in five lines
Twelve words on a durable medium, restore-tested before a single satoshi arrives. No
passphrase unless you have a concrete plan for storing a second secret for a decade,
in which case store it somewhere the seed is not.
A long, unique wallet password, understood as file protection rather than a backup.
Sweep loose keys instead of importing them, and guard your xpub as private data.
Raise the gap limit before you conclude anything is missing, and rehearse on testnet
before your first real transaction. Then read
the security model and decide whether your
balance justifies a hardware signer.
Two-line seed rule
Your 12 or 24 recovery words are the wallet. Write them on paper, never in a photo, cloud note or password manager, and never type them into a website, chat window or "support" agent — nobody legitimate will ever ask for them.
Setup questions people actually ask
Does Sparrow Wallet support a BIP39 passphrase?
Yes. Sparrow works with BIP39 mnemonic seeds and you can attach an optional passphrase, often called the "13th word" or "25th word". Remember what it actually does: the passphrase is mixed with the words to derive the keys, so every distinct passphrase yields a completely separate wallet from the same seed. Enter it differently and you get an empty wallet, not an error.
Is a 24 word seed phrase safer than 12 words in Sparrow?
Both are standard BIP39 mnemonics and Sparrow handles either. Twenty-four words carry 256 bits of entropy against 128 for twelve, but 128 bits is already unreachable by brute force, so the security gain is theoretical. The extra twelve words do double the amount you must transcribe and verify correctly, which is where real losses come from. Twelve words, backed up carefully and restore-tested, beats twenty-four words copied sloppily.
Does Sparrow Wallet support SLIP-39 shares?
Sparrow’s documented seed support is BIP39 mnemonics, and version 2.4.0 added Codex32 support. SLIP-39 share sets — the Shamir scheme used by some Trezor devices — are not part of that documented seed support, so do not plan a SLIP-39 recovery around Sparrow. If your goal is a backup that no single location can compromise, use multisig instead: the redundancy lives in the wallet policy rather than in a proprietary share format.
I restored my wallet and the balance is wrong. Where did my bitcoin go?
Usually nowhere. A restored wallet scans a limited run of unused addresses — the gap limit, 20 by default (40 for postmix wallets) — and stops when it hits that many empty ones in a row. If your coins sit past the gap, they exist on-chain but are invisible to the wallet. Raise the gap limit in the wallet settings, rescan, and check again before assuming loss. Also confirm your server is fully synced — see connecting a node.
What is the difference between the wallet password and the seed phrase?
The password unlocks a file on one computer. The seed words are the money. Losing the password costs you that copy of the wallet file, which you can rebuild from the seed; losing the seed words costs you the coins permanently, with no reset link and no support desk that can help.
Can I import a private key into Sparrow, or should I sweep it?
Sweep it. Sweeping means spending the coins from that key into an address your own seed controls — one on-chain transaction, and from then on your normal backup covers everything. Importing keeps you spending from a key that your seed does not derive, so your recovery story silently depends on a second secret you will forget to back up.
How do I try Sparrow without risking real bitcoin?
Use Tools > Restart in Testnet. Sparrow supports mainnet, testnet, signet and regtest, testnet coins are free from faucets, and rehearsing a full send-and-confirm cycle before your first real transaction is the cheapest insurance in self-custody.
