Level 1 · Start hereThe village#055

The 51% attack explained: outwrite the whole village, and you can unspend your own money

In a 51% attack, someone controls more mining power than the rest of the network and secretly builds a longer chain that rewrites recent history. They can reverse their own payments, not steal yours, and it has happened to smaller blockchains.

Ben buys a cart of firewood from Anna and pays with a line in the notebook. The villagers copy it. Anna delivers the wood. Nothing unusual yet.

But Ben has a plan. In his cellar, he has been writing his own version of the notebook, page by page, without the line paying Anna. The village has one rule for disagreements: the notebook with the most pages wins. If Ben can write pages faster than everyone else put together, his cellar copy will one day be longer. He'll bring it up, the village will switch to it, and the payment to Anna will be gone. Ben keeps the wood and the coins.

That's a 51% attack. It doesn't need anyone's password. It needs more writing power than the rest of the village combined.

What it is

On a proof-of-work blockchain, nodes follow the chain with the most work behind it. The Bitcoin developer guide says nodes "always follow the most difficult chain to recreate" (Bitcoin developer guide). Normally, that's the honest chain, because honest miners together have more computing power than anyone else.

The white paper bets on exactly that: "If a majority of CPU power is controlled by honest nodes, the honest chain will grow the fastest" (Nakamoto, 2008). A 51% attack is what happens when that bet fails. Someone with the majority mines a secret chain and then publishes it, replacing recent blocks. This is called a reorganisation.

How it works: what the white paper says, and what happened

Section 11 of the white paper looks at "the scenario of an attacker trying to generate an alternate chain faster than the honest chain". Its answers, step by step:

  1. What the attacker can't do. Even with a faster chain, the system isn't open to anything "such as creating value out of thin air or taking money that never belonged to the attacker". Why? "Nodes are not going to accept an invalid transaction as payment, and honest nodes will never accept a block containing them" (Nakamoto, 2008). Ben can't spend your coins, because he doesn't have your key.
  2. What the attacker can do. Reverse his own payments. Pay an exchange, wait for the coins to be credited, sell them or withdraw something else, then publish a chain where the payment never happened. That's a double spend.
  3. Why waiting helps. For an attacker with less than half the power, the paper says "the probability drops exponentially" with every block he has to catch up. Its table: against an attacker with 10% of the power, waiting 5 blocks gets the risk below 0.1%. Against 30%, you need 24 blocks.
  4. Why majority changes everything. With more than half, the attacker eventually catches up whatever you wait. Confirmations only buy time.

Now the real thing. Ethereum Classic, the smaller chain left behind by Ethereum's 2016 split, was attacked at least twice in the summer of 2020. Two of them are described by ETC Labs' chief executive, Terry Culver (ETC Labs, 2020):

  1. 31 July to 1 August 2020. The first attack "introduced approximately 3,000 blocks to the chain", with about $5.8 million stolen.
  2. 6 August 2020. The second "inserted 4,000 blocks", with a theft of about $1.7 million.
  3. The response. The project asked exchanges to "suspend deposits and withdrawals", and announced monitoring, research into a new mining algorithm, and work with miners and pools to raise the network's computing power.

Thousands of blocks of history, rewritten. Every one followed the rules. There were simply more of them.

Why it matters

  • Size is security. An attack needs more power than the honest network. On 18 November 2026, mempool.space estimated Bitcoin's hash rate at about 986 billion billion guesses per second (mempool.space). A small chain using the same kind of mining has far less to beat.
  • Exchanges carry the first risk. They're where an attacker can turn a reversed payment into something else. That's why they wait for many confirmations on small chains, and sometimes stop deposits.
  • Your coins aren't taken, but your payment can be. If you sold something and got paid in a coin that's then rewritten, your payment disappears with the old chain.
  • Proof-of-stake chains face a cousin of the problem. There, the majority is measured in staked coins, not computing power.

Where people get it wrong

  • "A 51% attack lets hackers steal everyone's coins." Not without the keys. The white paper rules that out; the attacker can only reverse his own spending.
  • "Bitcoin can't be 51% attacked." In theory it can. In practice, the attacker would need more computing power than the rest of the Bitcoin network combined.
  • "It needs exactly 51%." More than half wins eventually. Less than half can still get lucky over a few blocks, which is why confirmations matter.
  • "Pools with a large share are attacking." A large pool isn't an attack. An attack is a secret chain published to rewrite history. We cover pool shares in "What does decentralised mean?".

Wait for the pages to pile up

If you ever accept a payment in a small coin, don't hand over the goods at one confirmation. Find out how many confirmations exchanges ask for on that chain, and wait at least as long.

Ben's cellar notebook only works if Anna is in a hurry. Anna never is.

Frequently asked questions

What is a 51% attack?

It's when one person or group controls more than half of a blockchain's mining power and uses it to build a secret chain that replaces recent blocks. This lets them reverse their own transactions and spend the same coins twice.

Can a 51% attack steal my crypto?

Not directly. The Bitcoin white paper explains that an attacker can't create coins from nothing or take coins that never belonged to them, because nodes reject invalid transactions. The risk is to people who received a payment from the attacker that later disappears.

Has a 51% attack ever happened?

Yes, on smaller blockchains. Ethereum Classic was attacked at least twice in summer 2020, including on 31 July to 1 August and on 6 August. ETC Labs said thousands of blocks were reorganised and about $7.5 million was stolen across those two attacks.

How do confirmations protect against a 51% attack?

Each confirmation is another block an attacker must outpace. The white paper shows that against an attacker with less than half the power, the chance of success falls exponentially with each block. Against a majority attacker, confirmations only delay the attack.

Is Bitcoin safe from a 51% attack?

No network is immune, but Bitcoin has the largest proof-of-work network, which makes an attack very costly. An attacker would need more computing power than all the other Bitcoin miners combined.

Sources

  1. Satoshi Nakamoto, Bitcoin: A peer-to-peer electronic cash system (2008), sections 4 and 11
  2. Bitcoin developer guide, Block chain
  3. Terry Culver (ETC Labs), EDCON 2020: Ethereum Classic resilience
  4. mempool.space API, bitcoin hash rate (read on 18 November 2026)

Read next.