Guide
What Is a Bitcoin 51% Attack? Does It Actually Happen, and What Can — and Can't — It Do

Bottom line
A Bitcoin 51% attack is when an attacker who controls a majority of the network's total mining power (hashrate) rewrites recent blocks to their own advantage. This makes it possible to "cancel a transaction you already sent and spend the same coin again" (double-spending) and to "censor" specific transactions by refusing to confirm them. But it is absolutely not possible to steal other people's wallet balances, forge signatures, or rewrite the 21 million coin supply cap. And against Bitcoin itself, the scale of hardware and electricity required is so unrealistic that it has never succeeded. Real-world damage from this type of attack has hit other, smaller-hashrate proof-of-work chains instead.
Key points of this article
- What it can do: "double-spend" transactions the attacker was involved in, and "censor" (refuse to confirm) specific transactions
- What it can't do: steal other people's balances, forge signatures, or alter the coin issuance cap or deeply buried transaction history
- The cost involved: attacking Bitcoin requires industrial-scale ASIC hardware and enormous electricity — practically unobtainable
- Where it has actually happened: not on Bitcoin itself, but on smaller chains like Ethereum Classic and Bitcoin Gold
What a 51% attack actually is
Bitcoin runs on a rule where miners around the world race to add blocks via a computational competition, and the longest accumulated chain is treated as the correct record (explained in more detail in how Bitcoin works).
If a single attacker controls more than half of the network's total computing power, they can build blocks faster than everyone else combined. That lets them secretly build an alternative, longer chain and swap it in for the one that was already public, effectively erasing recent transaction history. That's a 51% attack. "51%" is shorthand for a majority — strictly, holding a stable majority above 50% is what makes the attack possible.
The core resource behind the attack is "hashrate" — a measure of mining power. Whether the attack is even feasible ultimately comes down to "what percentage of total hashrate can the attacker muster?" For the basics of mining, see what is mining.
What an attacker CAN do with 51%
An attacker holding a majority can mainly do two things:
1. Double-spending The attacker first sends their own coins to, say, an exchange, and converts and withdraws them into fiat currency or another coin. Behind the scenes, they secretly mine an alternate chain that excludes that transaction, then later publish it to replace the original. The "record of having sent the payment" disappears, and the coin effectively returns to the attacker's hands — they've spent the same coin twice. This is the primary goal of a 51% attack.
2. Censoring transactions By excluding specific transactions from the blocks they build, an attacker can obstruct those transactions' confirmation. This is delay and obstruction, though — it does not move anyone else's assets.
What an attacker CANNOT do (the most important part)
Most of the fear around this topic comes from misunderstanding this part. Even with a majority of hashrate, an attacker can absolutely never:
- Steal balances from other people's wallets — sending requires a signature from the owner's private key, and majority hashrate does not let you generate someone else's key
- Forge signatures or impersonate someone to send their funds — cryptographic signing is a separate problem from raw computing power
- Increase the 21 million coin supply cap or mint coins out of thin air — every node's validation rules would reject it
- Rewrite deeply buried, old transaction history — the deeper a block is buried, the more astronomically difficult it becomes to reverse, to the point of being practically impossible
In other words, a 51% attack can only "reverse recent transactions the attacker themselves was involved in." Coins you hold and control with your own keys cannot be directly seized through this kind of attack.
| Capability | Possible | Not possible |
|---|---|---|
| Reversing your own transaction (double-spend) | Yes | — |
| Censoring / blocking confirmation of specific transactions | Yes | — |
| Stealing other people's balances | — | Yes |
| Forging signatures / impersonated transactions | — | Yes |
| Changing the supply cap or coin issuance | — | Yes |
| Rewriting old transaction history | — | Yes (practically impossible) |
How expensive would it actually be to attack Bitcoin
Some websites that model theoretical attack costs (such as crypto51.app) sometimes show a one-hour attack costing in the millions of dollars. But that figure rests on an unrealistic assumption — that the necessary hashrate could actually be rented on the open market.
The real barrier isn't price, it's supply. Bitcoin's total hashrate in recent years has reached roughly the hundreds of EH/s to 1 ZH/s range (check difficulty adjustment and current hashrate statistics for the latest figures), and no rental market comes close to supplying enough dedicated SHA-256 ASICs to command a majority of that. An attacker would need to build industrial-scale mining facilities themselves and continuously consume electricity on the scale of a city. And even if they somehow pulled it off, the moment the attack became known, bitcoin's price would likely crash — destroying the value of the very hardware they invested in. In short, it doesn't make economic sense. This is why Bitcoin is considered effectively secure against 51% attacks.
It has happened on smaller chains
By contrast, on proof-of-work chains with much smaller hashrate, an attacker can rent the needed computing power on the open market — and real damage has occurred.
- Ethereum Classic (ETC): In January 2019, a chain reorganization enabled a double-spend of roughly 219,500 ETC (worth about $1.1 million at the time). In August 2020, multiple attacks involving reorganizations of thousands of blocks occurred, with reported damages in the millions of dollars.
- Bitcoin Gold (BTG): A large double-spend attack in May 2018 (reported at roughly $18 million), and another deep-reorganization double-spend in January 2020.
The root cause in both cases was that these chains' hashrate was small enough to be rentable. It's a clear illustration that even under the same proof-of-work design, the scale of hashrate determines security.
注意
This article is educational, explaining how 51% attacks work — it is not investment advice. Exchanges defend against attack risk by requiring more confirmations. During the Ethereum Classic attacks, major exchanges sharply raised their required confirmation counts. For large deposits and withdrawals, waiting for the number of confirmations your exchange specifies is the basic practical defense against double-spending.
Frequently asked questions
Q. Has a 51% attack ever succeeded against Bitcoin? A. Never. Commanding a majority of hashrate requires industrial-scale hardware and enormous electricity, and it's not economically worthwhile. Real-world damage has occurred on smaller-hashrate chains, not Bitcoin itself.
Q. If a 51% attack happens, will my bitcoin be stolen? A. No. An attacker cannot generate someone else's private key, so they cannot directly seize coins you hold. The only thing the attack enables is reversing recent transactions the attacker themselves was involved in.
Q. Why is it called "51%" specifically? Wouldn't 50.1% not be enough? A. "51%" is shorthand for a majority. Strictly, holding a stable majority above 50% is what lets an attacker build blocks faster than everyone else combined — and the more comfortably over 50%, the higher the success rate.
Q. How do you protect against double-spending? A. By having exchanges and recipients wait for a sufficient number of confirmations. The deeper a block is buried, the exponentially harder it becomes to reverse.
Sources
- Bitcoin.com — What is a 51% Attack?
- Braiins — How Much Would it Cost to 51% Attack Bitcoin?
- Crypto51 — Cost of a 51% Attack for Different Cryptocurrencies
- CoinDesk — Ethereum Classic Suffers Second 51% Attack in a Week (2020)
- Cointelegraph — Bitcoin Gold Blockchain Hit by 51% Attack (2020)
A note before investing
This article is intended for informational purposes and is not investment or tax advice. Bitcoin carries risks including price volatility, hacking, and loss. Make investment decisions at your own responsibility and only with funds you can afford to risk. This article reflects publicly available information as of its publish date; always verify current details with primary sources.
Sources
- Bitcoin.com — What is a 51% Attack?
- Braiins — How Much Would it Cost to 51% Attack Bitcoin?
- Crypto51 — Cost of a 51% Attack for Different Cryptocurrencies
- CoinDesk — Ethereum Classic Suffers Second 51% Attack in a Week
- Cointelegraph — Bitcoin Gold Blockchain Hit by 51% Attack Leading to $70K Double Spend
FAQ
- Has a 51% attack ever succeeded against Bitcoin?
- Never. Commanding a majority of hashrate requires industrial-scale ASIC hardware and enormous amounts of electricity, and it's not economically worthwhile. Real-world damage has occurred on smaller-hashrate PoW chains like Ethereum Classic and Bitcoin Gold, not Bitcoin itself.
- If a 51% attack happens, will my bitcoin be stolen?
- No. An attacker cannot generate someone else's private key, so they cannot directly seize coins you hold. What the attack enables is reversing recent transactions the attacker themselves was involved in (double-spending) and censoring specific transactions — nothing more.
- Why is it called "51%" specifically?
- "51%" is shorthand for a majority. Strictly, holding a stable majority above 50% is what lets an attacker build blocks faster than everyone else combined, so the attack becomes possible — and the more comfortably over 50%, the higher the success rate.
- How do you protect against double-spending?
- By having exchanges and recipients wait for a sufficient number of confirmations. The deeper a block is buried under later blocks, the exponentially harder it becomes to reverse. For large deposits or withdrawals, always wait for the number of confirmations your exchange requires.
This article is informational only and is not financial, investment, or trading advice. Prices are reference snapshots and may be outdated. Always do your own research.