Node: What Blockchain Nodes Do and Which Types Exist
A node, also called a Bitcoin node or blockchain node, is a computer that connects to the Bitcoin network and exchanges transactions and blocks with other nodes, its peers. The network is peer-to-peer (P2P), which means the computers connect directly to one another and pass data between themselves. Only full nodes check every block against the consensus rules.
What Does a Bitcoin Node Do on the Peer-to-Peer Network?
A Bitcoin node runs Bitcoin software on a computer and keeps that computer connected to its peers. Bitcoin is a cryptocurrency, and its transactions and blocks travel over these connections between nodes.
A Bitcoin node passes on what it receives. A transaction is a record that moves BTC from one owner to another, and a block is a batch of transactions that is added to the blockchain together. When a node receives a new transaction or a new block, it can pass the item on to its own peers.
A Bitcoin node connects to some peers, not to every other node on the network. Transactions and blocks spread from peer to peer, and two nodes can hold different information at the same moment.
The blockchain is the shared record of all blocks, in order. Every block has a header, which is the summary data at the front of the block. The header holds the hash of the previous block’s header. A hash is a fixed-length fingerprint calculated from data. The chain is designed so that changing a transaction changes the hash of its block, and the changed hash no longer matches the one stored in the next block.
Which Types of Nodes Exist, and What Does Each One Check?
Nodes differ in how much they check and how much data they keep.
What Is a Full Node?
A full node is a node that downloads every block, checks every block and every transaction in it against the consensus rules, and passes on the valid ones. The consensus rules are the rules that every full node applies to decide what is valid, so that full nodes stay in agreement with each other.
A full node keeps its own chain, made only of blocks it has validated, and it takes no other node’s word that a block is valid. A full node that stores the entire blockchain is also called an archival node.
What Is a Pruned Node?
A pruned node is a full node that deletes old block data after it has validated that data. It checks every block and transaction as a full node does. It keeps the recent blocks and the set of unspent transaction outputs (UTXO set), because it needs that set to check new transactions. An unspent transaction output is an amount of BTC recorded on the blockchain that has not yet been spent.
A pruned node needs less disk space, and it serves only recent blocks to other nodes, because it no longer holds the older ones.
What Is a Light Client (Light Node or SPV Client)?
A light client is software that uses the Bitcoin network without downloading and checking every block and transaction. It is also called an SPV client, for simplified payment verification, a thin client, a lightweight client, a lightweight node, or a light node. Bitcoin’s developer documentation separates it from a node, because a light client does not relay blocks and transactions and does not check them all.
A light client downloads block headers, which are 80 bytes each, and the data needed to check a payment. To check a payment, a light client asks a full node for a Merkle branch, a short chain of hashes that links one transaction to the block that holds it. With the branch and the headers, a light client can check that a transaction is included in a block.
A light client cannot check that the transaction follows every rule, so it relies on full nodes for proofs and on the chain with the most work, which is the chain that would take the most effort to recreate. The method holds while honest participants control most of the network’s computing power, and an attacker who overpowers the network could fool it.
What Does a Full Node Do with a Transaction and with a Block?
A full node receives transactions and blocks from its peers and typically does the following with each.
When a transaction arrives, a full node goes through these steps:
- The inputs are checked. Each input must refer to an output that is still unspent, and its data, typically a digital signature, must satisfy the spending condition of the output it uses.
- The amounts are checked, because the outputs must not add up to more than the inputs.
- The node may keep a valid transaction in its mempool, the waiting area where a node holds transactions that are not yet in a block, and pass it on to its peers. Whether it does so depends on the node’s relay policy, a set of extra local rules. A transaction that fails a check is dropped.
A worked example with made-up numbers shows the amounts check. In it, two transactions, X and Y, each spend one input of 0.5 BTC. Transaction X sends 0.3 BTC to the receiver and 0.1999 BTC back to the sender as change, which is the part of the input that the sender does not spend on the payment. Transaction Y sends 0.3 BTC to the receiver and 0.2001 BTC back as change.
The amounts in this example are made up and are not a fee recommendation, and a real full node also checks that each input is still unspent and that each signature is valid, which the example does not show.
A block that arrives from a peer goes through these checks, among others:
- The proof of work is checked. The hash of the block header must be at or below the target, a number set by the network, and proof of work is the computing effort miners spend to find such a hash.
- The block’s structure is checked. The Merkle root, a hash in the header that is calculated from the block’s transactions, must match those transactions, and the block’s weight, a measure of its size, must stay within the limit that the consensus rules set.
- The transactions are checked. Every transaction other than the first must pass the transaction checks above. The first transaction, called the coinbase transaction, has no input to check, and it may claim no more than the block subsidy, the newly created BTC in a block’s reward, plus the transaction fees in that block.
- A valid block is added to the node’s chain. The node typically removes the transactions that the block contains from its mempool and announces the block to its peers.
A full node rejects a block that breaks any of these rules, such as a coinbase transaction that claims more than the subsidy plus the fees.
The block subsidy is 3.125 BTC per block as of October 2026, in force since April 2024, and it is cut in half at each halving.
Why Does It Matter That Each Full Node Checks the Rules Itself?
A full node applies the consensus rules to every block on its own. It does not rely on a peer, a miner, or a company to say that a block is valid.
A full node enforces a change to the consensus rules only if its operator runs software that contains the change. After a soft fork, one kind of rule change, a node that has not upgraded keeps accepting the same blocks as upgraded nodes, as long as most of the hashrate, the combined computing power miners add to the network, enforces the new rule, but it does not check the new rule itself. After a hard fork, another kind, a node that has not upgraded can be unable to validate blocks made under the newer rules.
Full nodes are designed to follow the chain with the most work. A party that holds most of the hashrate could revise recent blocks. The rules that full nodes enforce still apply in that case, so a block that creates more BTC than the rules allow, or spends BTC without a valid signature, is designed to be rejected by nodes that check.
How Is a Node Different from a Miner?
A miner and a node do different jobs. A miner’s block is assembled from waiting transactions, by the miner or by the pool the miner works for, and the miner spends computing effort on the proof of work. A node checks the finished block and the transactions in it, and node software does not mine unless the operator also runs mining software.
One operator can both mine and run a node. A solo miner, who mines alone and not through a pool, typically uses a full node’s software to get waiting transactions and to broadcast the block it finds.
What Does Running a Node Involve?
Running a full node means running node software, such as Bitcoin Core, that connects to peers, downloads the blockchain, checks it, and keeps checking each new block and transaction as it arrives.
A new full node first has to download and check every block from the first block, called the genesis block, up to the latest one. This catch-up is called the initial block download, and how long it takes depends on the computer and the connection.
A full node needs disk space to store block data and bandwidth to download and relay it. The software’s own documentation lists hardware needs, setup steps, and port numbers, and those change as the blockchain grows and the software changes.
Sending and receiving BTC does not require running a node, because a wallet can work as a light client and rely on other people’s nodes.
What Does a Node Not Do?
A Bitcoin node does not hold other people’s BTC. The blockchain records which outputs exist, and the private keys that can spend them are kept in wallets. A private key is a secret number that lets its owner authorize a spend.
A node that accepts or relays a transaction does not confirm it. A transaction is confirmed once a block includes it, and Bitcoin transactions are designed to become harder to reverse with each block added after that one. Until then, an unconfirmed transaction can still be replaced or dropped.
In Bitcoin, the block subsidy and the transaction fees go to the miner whose block is added to the chain, and the first transaction in that block collects them.
November 27, 2025
