A replay attack occurs when an attacker reuses a previously captured and valid message, signature, or transaction to cause an operation to be executed again.

A digital signature proves that a particular set of data was signed by the corresponding key, but a signature alone does not necessarily guarantee that the message can only be used once. A protocol must include context, uniqueness, and state information to prevent unintended reuse.

Blockchain replay protection may use transaction inputs, nonces, network identifiers, chain identifiers, or other fields. These mechanisms distinguish a new operation from one that has already been processed.

Replay risks are particularly well known when one blockchain splits into two networks after a hard fork. If the transaction formats and signature rules do not sufficiently distinguish the chains, a transaction valid on one network may potentially be replayed on the other.

Replay protection is therefore part of transaction-format design, digital-signature design, and the rules of the specific blockchain.