What is a UTXO? How Bitcoin keeps track of coins without any accounts

Think banknotes in a drawer, not a balance at a bank

A bank keeps one number per customer and adjusts it. Bitcoin does not. It keeps a list of outputs: lumps of value, each created by some earlier transaction and each locked with a condition that says who may spend it. An output that nobody has spent yet is an unspent transaction output, a UTXO. The Bitcoin developer guide puts the consequence plainly: “When your Bitcoin wallet tells you that you have a 10,000 satoshi balance, it really means that you have 10,000 satoshis waiting in one or more UTXOs.”

The closest everyday picture is cash. If you have a $20 note and two $5 notes, you have $30, but you do not have a $30 note. Each note is a separate object, spendable only whole. Bitcoin outputs are the same, except that they can be any size (measured in satoshis, the smallest unit) and the “note” is locked to a key rather than sitting in a drawer.

The white paper starts from this picture. Its second section opens: “We define an electronic coin as a chain of digital signatures.” Each owner passes the coin on by signing the previous transaction and the next owner’s public key. A “coin” in that sentence is what we now call an output, and its history is the chain of transactions that created and spent it. The protocol never keeps a table of who owns how much; block explorers compute a balance for an address by adding up the unspent outputs locked to it.

A transaction consumes whole outputs and creates new ones

A transaction is a small record with two lists. Its inputs point at existing UTXOs; the developer reference describes each input as naming “the TXID of the transaction holding the output to spend” and “the output index number of the specific output to spend,” a pair often written as txid:vout. Its outputs create new UTXOs, each with a value in satoshis and a script setting the condition for spending it. Nothing is edited in place: the old outputs are consumed and the new ones appear.

Two rules make it work. The reference states the first: “the sum of all outputs may not exceed the sum of satoshis previously spent to the outpoints provided in the input section.” You cannot create value. The developer guide states the second: because “a UTXO can only be spent once, the full value of the included UTXOs must be spent.” Like the $20 note, an output is consumed entirely, and any surplus has to be sent somewhere deliberately.

Section 9 of the white paper anticipates exactly this. “Although it would be possible to handle coins individually, it would be unwieldy to make a separate transaction for every cent in a transfer. To allow value to be split and combined, transactions contain multiple inputs and outputs. Normally there will be either a single input from a larger previous transaction or multiple inputs combining smaller amounts, and at most two outputs: one for the payment, and one returning the change, if any, back to the sender.”

Account for the recipient, change and fee

Suppose a payment spends a 100,000-satoshi output, creates 60,000 satoshis for the recipient and pays a 1,000-satoshi fee. The remaining 39,000 satoshis can become a new change output controlled by the sender. The old output is consumed in full.

The fee is the difference between inputs and outputs, not a separate fee destination. An explorer can show the new outputs without knowing which one is the sender’s change. The wallet’s ownership information provides that context.

Every output traces back to a block reward

Follow any output backwards through the transactions that created it and you reach a transaction without consuming a previous UTXO at all. The white paper describes it in section 6: “By convention, the first transaction in a block is a special transaction that starts a new coin owned by the creator of the block.” That is the coinbase transaction, the only place new bitcoin comes from. The developer reference shows how the format marks it: its single input carries a 32-byte field of zeros where a previous transaction ID would be, and an index of all ones.

Coinbase outputs come with a waiting period. The developer guide notes that such an output “cannot be spent (used as an input) for at least 100 blocks.” The reason is that a freshly mined block can still be displaced if the network settles on a competing block, and a reward that vanished after being spent would unravel every transaction built on it. Our article on confirmations explains why a block’s place in the chain firms up over time.

The very first coinbase output, from block 0, is in the site’s wallet library under the genesis block. New issuance enters through coinbase outputs, but a transaction with multiple inputs and outputs does not assign particular input satoshis to particular outputs. Graph paths are not proof of literal coin ancestry.

The UTXO set is the ledger a node actually consults

A full node does not need to re-read the whole history to decide whether a new transaction is valid. It keeps the current set of unspent outputs as a database and checks each input against it: is this output in the set, and does the input satisfy its lock? When a block is accepted, the spent outputs are removed and the new ones added. Bitcoin Core exposes the state of this database through a command called gettxoutsetinfo, which the reference describes as returning “statistics about the unspent transaction output set,” including the block height, “the number of unspent transaction outputs,” the number of transactions that still have unspent outputs, and the total amount they hold.

The white paper foresaw this in section 7: “Once the latest transaction in a coin is buried under enough blocks, the spent transactions before it can be discarded to save disk space.” What can never be discarded is the set of outputs still unspent, because that is what defines who can spend what.

This state also enforces the single-spend rule. Two transactions cannot both consume the same output in one accepted chain. Input availability and chain ordering determine which spend is accepted; being first seen by a wallet or explorer is not a consensus priority claim. Our double spending article follows a transaction from input selection to confirmation and possible reorganization.