Immutability is the property that, once a transaction is confirmed on a blockchain and has enough blocks built on top of it, it cannot be edited, reversed, or deleted. The chain is append-only. You can write new transactions that undo the economic effects of old ones — sending tokens back to the original sender, for example — but the original transaction remains in the historical record forever, visible to anyone looking at a block explorer. This is what makes blockchain audit trails different from traditional databases: there is no “update” or “delete” operation at the protocol level, only “append”.
The mechanism is not magic. Immutability comes from the combination of cryptographic hash linking and economic cost. Every block contains the hash of its parent, so changing a transaction in a historical block would change that block’s hash, which would break the next block’s reference, which would break the one after that, and so on. To rewrite history, an attacker would need to redo the proof-of-work (or get the stake weight) for every block from the change forwards and then outpace the honest network. For a chain like Bitcoin with a decade of accumulated work behind its current tip, this is economically impossible in any realistic scenario, which is why Bitcoin transactions with a few confirmations are considered effectively final.
What Immutability Actually Gets You
The value of immutability is that it makes certain kinds of trust cheaper. Two parties who do not trust each other can share a ledger with confidence that neither can unilaterally alter past entries. An auditor checking the chain years later can know they are seeing the same data that everyone else saw. A smart contract enforcing a rule based on past transactions can rely on those transactions not being silently rewritten underneath it. These properties are hard to get from centralised databases without substantial legal and institutional scaffolding — trusted third parties, notarisation, backup systems, audit logs — and a blockchain gets them from the protocol itself.
For public record-keeping, provenance tracking, and anywhere the value proposition is “multiple parties need to see the same data and none of them want the others to be able to change it”, immutability is a real feature. For private or intra-company databases where all parties trust each other, a blockchain is usually overkill and a regular database works fine.
The Downside
Immutability cuts the other way when something goes wrong. If you send crypto to the wrong address, it is gone. If you sign a transaction that approves a malicious contract to spend your tokens, the approval is on the chain and cannot be undone. If a smart contract has a bug that lets an attacker drain funds, the drain transactions are permanent. The DAO hack of 2016, where an attacker exploited a reentrancy bug to steal 3.6 million ETH, is the most famous example of this tension — the Ethereum community eventually chose to hard-fork the chain to reverse the hack, which worked but arguably broke the immutability promise the chain was supposed to provide. The remaining original chain became Ethereum Classic, which retained the hack as a statement of principle.
This tradeoff is permanent and unresolved. Every time a major exploit happens on a chain, there is a debate about whether the community should fork to undo it, and the answer depends on whether you value the abstract principle of immutability more than the concrete outcome of not losing money. Most of the time, the answer is “do not fork”, and victims have to absorb the loss.
How Immutable Is “Immutable”?
There are edge cases. Chain reorganisations — where a block that seemed confirmed gets replaced by a competing fork — are normal at shallow depths on proof-of-work chains, which is why exchanges typically wait six confirmations for Bitcoin deposits. Proof-of-stake chains with slashing and finality gadgets (Ethereum post-Merge) provide stronger guarantees: once a block is finalised, reverting it would require a supermajority of stake to get slashed, which makes deep reorgs economically prohibitive in a different way. The result is that “immutable” is better thought of as “increasingly expensive to change, to the point where nobody will” rather than as a hard mathematical guarantee.