Merkle Block

A Merkle block is a Bitcoin peer-to-peer message (defined in BIP-37) that a full node sends to an SPV client. It contains:

• The full block header (80 bytes).
• The hashes of the transactions in the block that match the SPV client's Bloom filter.
• A Merkle proof (a partial Merkle tree) showing those transactions are committed in the block's Merkle root.

The matching transactions themselves are not in the message. The full node sends them right after it, as ordinary tx messages.

This is what lets a phone wallet check whether its transactions appeared in the latest block, without downloading the whole block.

The downside, which has gotten more attention over the years: Bloom filters leak privacy. The SPV client sends its filter to the full node, which can probabilistically reverse-engineer which addresses or transactions the client cares about. For a privacy-conscious user, this is a meaningful concern. BIP-37, written in 2012, expected clients to hide their addresses by choosing a looser filter, but a 2014 study found that a wallet using fewer than 20 addresses could reveal almost all of them.

The modern replacement is BIP-157/158 compact block filters: the server computes a filter per block, the client downloads it and checks for matches locally without revealing which addresses are theirs. Better privacy at modest bandwidth cost. Some mobile wallets use this approach rather than BIP-37 Merkle blocks.

As of October 2026, Bitcoin Core still implements merkle blocks. Since version 0.19.0.1 (November 2019), though, it serves them only when the node's operator turns Bloom filter support back on. The release notes gave denial-of-service risk as the reason.