← All guides

Multisig wallets explained: when one key is not enough

Security · 9 min read

Financial disclaimer, not financial advice. This article is general information and education only. It does not recommend buying, selling, or holding any crypto asset, fund, or product, and it is not personalised financial, tax, or legal advice. Crypto assets are volatile and speculative and can lose all of their value. Consider your own circumstances and, where appropriate, speak with a qualified professional in your jurisdiction before making any financial decision.

How many keys should have to agree before money can move? A single-key wallet answers “one”, one key signs, and the transaction goes. A multisig wallet raises the number: a set of keys exists, and a stated threshold of them must sign before anything moves. Raising that number removes one failure, a single lost or taken key emptying the wallet, and creates another: lose enough keys, and no one can sign at all. Which side of that trade matters more depends on the holder’s situation, not on the technology. Start with the side most explainers leave for last.

First, the stop line: when multisig is the wrong tool

Do not add a threshold to solve a problem a single key does not have.

If one person alone controls the funds, the amount at risk is small relative to the cost of managing several keys, and no second person legitimately needs a say in a transaction, multisig adds signers, devices, and recovery paths without removing a real single point of failure. The complexity itself becomes the risk: more keys to store, more keys to lose, and a threshold that can lock the holder out of their own funds.

The stop line runs the other way as well. If losing one key, or having one key-holder unavailable, pressured, or gone, would strand or expose the whole amount, and the amount justifies the setup and ongoing care, staying on one key because it is simpler is its own decision, with its own cost. The sections below give the mechanics needed to place a given situation on one side of that line or the other.

What m-of-n actually enforces, on-chain

Multisig is short for multiple signatures. A wallet is set up with n keys and a threshold m: any m of the n keys can authorise a transaction, and fewer than m cannot. Two configurations dominate the examples. A 2-of-3 setup is popular for personal security; a 3-of-5 setup is common for organisations.

Until the threshold is met, nothing reaches the blockchain. A transaction becomes valid only once the required number of signatures is present; a partly signed transaction is not a smaller transaction, or a pending one with partial effect. It is not yet a transaction at all.

The enforcement lives in different places on different networks. Multisignature technology has been part of Bitcoin since 2012. On Bitcoin, a multisig address is derived from the set of public keys and the threshold, using address forms identified as P2SH and P2WSH. On Ethereum, the same idea is usually carried by a smart-contract wallet. Safe is the standard named example, where the contract itself holds the list of owners and enforces the threshold. Casa and Unchained are also named in the multisig landscape, on the service side of the setup rather than inside the address or the contract.

That placement matters for one practical reason: the threshold is not a policy a provider promises to follow. On Bitcoin it is built into the address; on Ethereum it is built into the contract. Either way, a signer below the threshold cannot move funds by persuading a support desk.

Three failure modes, run against each setup

Abstract safety claims settle nothing. Run the same three failures against each configuration and count what is left.

One key is lost. A single-key wallet has no second key: the count of remaining signers falls to zero, below any threshold of one. A 2-of-3 wallet loses one key and keeps two, exactly the threshold, so funds remain movable, with no spare left. A 3-of-5 wallet loses one and keeps four, a margin of one above its threshold of three.

Keys keep being lost. This is the failure in the other direction, and it is permanent when it completes. A 2-of-3 wallet that loses a second key holds one signature against a threshold of two: the funds are locked, not by an attacker, but by the setup doing exactly what it was told to do. A 3-of-5 wallet reaches the same wall after a third loss. The margin worth naming at setup time is therefore n minus m, the number of keys that can be lost before signing becomes impossible. For 2-of-3 that margin is one. For 3-of-5 it is two.

One signer is unavailable, pressured, or gone. A threshold above one means a single signer cannot authorise a move alone, and a single missing signer does not block a move either, provided the remaining keys can still reach m. The same arithmetic governs both directions: count the keys that remain able and willing to sign, and compare that count with the threshold. Above it, the wallet works. Below it, it does not, whatever the reason for the shortfall.

SetupThresholdKeys that can be lost before lockoutOne signer alone can move funds?
Single key1 of 10Yes, that is the design
Multisig2 of 31No
Multisig3 of 52No

Read the middle column as the price of the last column. Every step that makes “one signer alone” harder also consumes margin against ordinary loss, accident, and time.

Multisig versus MPC: what can be verified on-chain

MPC wallets are often sold in the same conversation, and the distinction is technical, but it has a practical edge a holder can check. MPC, short for multi-party computation, splits one key across devices and produces a single signature. Multisig does not split one key. It records multiple signatures, or enforces a threshold inside a smart contract.

The consequence is visibility. A multisig threshold is an on-chain fact: the address or the contract states how many signatures a transaction needs, and a transaction that reaches the chain shows the threshold was met. An MPC arrangement produces one ordinary-looking signature, so the sharing arrangement behind it is not visible in the signature itself. Neither fact settles which approach suits a given holder. It does settle what a holder, an heir, or a co-signer can verify without trusting a provider’s description of its own system.

Collaborative custody: the middle option, with its condition attached

Between holding every key alone and handing funds to a custodian sits a shared arrangement: a provider holds one of the keys in a multisig set, and the holder keeps the others. Casa and Unchained are named examples of services in this space.

The condition that decides what the arrangement actually is can be stated in one sentence: collaborative custody stays non-custodial as long as the keys the holder controls can meet the threshold without the service. In a 2-of-3 set where the holder keeps two keys and the provider keeps one, the holder’s two signatures reach the threshold alone, so the provider’s participation is help, not permission. Reverse the split, holder keeps one, other parties keep the rest, and the holder cannot move funds unaided, whatever the arrangement is called.

Count the keys, not the label. That single count classifies the setup more reliably than any product name.

Changing the set later: rotation is a governed transaction

A key set is not necessarily fixed at setup. Some implementations let owners rotate keys or update the list of owners through a governed transaction, a change authorised under the wallet’s own threshold rules, rather than by moving all funds to a fresh wallet.

That capability belongs in the setup decision, because the failure it answers is predictable: a key is lost while the margin still allows signing, a signer leaves an organisation, or a device is replaced. A setup that can rotate a lost key while the remaining keys still meet the threshold turns a lost key into maintenance. A setup that cannot leaves the reduced margin in place, one loss closer to lockout, until the funds are moved. Ask which of those applies before the first loss, not after it.

The decision rule: when one key is not enough, and when it is

  1. Shared control is genuinely required. More than one person must be able to block or approve a move, a shared treasury, a partnership, an organisation using a 3-of-5 pattern. One key cannot express that requirement at all.
  2. One loss would be catastrophic and unrecoverable. The amount at risk is large enough that a single lost, taken, or unavailable key is an unacceptable single point of failure, and large enough to justify the setup cost and the ongoing care of several keys.
  3. Succession has to work without one person. If a signer dies or becomes unavailable, the remaining keys must still reach the threshold. Check that arithmetic explicitly: remaining keys against m, not against good intentions.
  4. The lockout margin is acceptable. State n minus m out loud. If losing that many keys plus one is a plausible story for the household or organisation involved, the chosen configuration, or the key-storage plan behind it, is wrong.
  5. None of the above applies. Then one key is enough, and the honest work is elsewhere: a careful backup of that key, stored well. Multisig would add failure modes faster than it removes them.

What a sound decision looks like: the threshold, the margin, and the keys each party holds can all be stated in numbers, and the lockout case has been counted rather than assumed away. A setup that can only be described in adjectives, safer, more secure, more serious, has not been decided yet.

, Sterling Cole, Money Writer. Earning Desk

Reminder: Crypto Investing is educational content only, not financial advice. Crypto assets are volatile and can lose all value. Product terms, fees, and rules change, read the current prospectus, terms, or manufacturer documentation for any product you are considering.