A transaction fee is a bid for space in a block, priced in satoshis per virtual byte, and it goes to the miner who includes your transaction. Fees and newly created coins together pay miners to keep Bitcoin’s history expensive to rewrite, and as of November 2026 fees supply well under 1% of that pay. You set a fee every time you send bitcoin, so knowing how the auction works saves you money and spares you stuck payments. And because the new-coin part halves about every four years, how fees grow decides how Bitcoin will pay for its own protection in the decades ahead.
A Fee Is a Bid for Block Space
A Bitcoin transaction has no “fee” field. It spends some coins (its inputs) and creates new ones (its outputs), and whatever the inputs hold beyond the outputs is the fee. The miner whose block includes the transaction claims it in the block’s first transaction, the coinbase transaction.
What a fee buys is room in a block, and that room is fixed: about one block every 10 minutes, each holding at most 4 million weight units, the measure Bitcoin’s rules use for block space. Wallets count in virtual bytes (vB); one vB is four weight units, so a full block holds about 1 million vB. The price of that room is the fee rate: the fee divided by the transaction’s size, in satoshis per virtual byte (sat/vB). A satoshi is one hundred-millionth of a bitcoin. The transactions lesson covers the details.
Why Size Sets the Fee, Not the Amount
Look at it from the miner’s side. Your transaction costs no extra electricity; it costs space that could have held someone else’s fee. A 141 vB transaction pushes out 141 vB of others whether it moves 5 BTC or 5,000 sats. So the auction prices space, and the amount plays no part. What makes a transaction big is mostly how many coins it spends and what kind of address they sit in:
| Payment | Inputs and outputs | Size | Fee at 10 sat/vB |
|---|---|---|---|
5 BTC from one coin in a native SegWit address (bc1q) | 1 in, 2 out | about 141 vB | 1,410 sats |
0.01 BTC from one coin in an old address starting with 1 | 1 in, 2 out | about 226 vB | 2,260 sats |
0.01 BTC gathered from ten small coins (bc1q) | 10 in, 2 out | about 753 vB | 7,530 sats |
The small payment built from ten coins costs more than five times the 5 BTC one: each extra bc1q input adds about 68 vB. The address types lesson compares the formats.
The auction is also pay-your-bid: each transaction pays exactly the fee written into it, even if a lower rate would have got it into the same block. Bidding far above everyone else only gives money away.
The Mempool Is a Queue Sorted by Bid
Transactions waiting to be mined sit in the mempool (memory pool). Each node keeps its own. The mempool is a queue, but not first come, first served: each newcomer steps in ahead of everyone bidding less. Read the queue in blocks. Every million vB waiting above your fee rate is roughly one more block, about ten minutes, before your turn, as long as nobody new outbids you.
On the morning of 6 November 2026, mempool.space showed about 17.7 million vB waiting, enough to fill about 18 blocks, and almost all of it bid between 0.1 and 0.2 sat/vB. That 0.1 sat/vB floor is the lowest rate Bitcoin Core, the most widely used node software, has passed on to other nodes by default since version 29.1 (September 2025). Almost every recent block was full, and yet fees were tiny. Block 974,789, mined by the Foundry USA pool, held 7,316 transactions that paid 140,639 sats in fees in total, about $107, next to a new-coin reward worth about $238,000.
Full blocks don’t mean high fees. A place in the next block costs what the cheapest transaction that still fits is paying. That price climbs only when more than a block’s worth of transactions bid well above the floor, so that people who need the next block have to outbid each other. When nearly everyone in the queue bids the minimum, as on that morning, a full block still sells for almost nothing.
How Miners Choose What Goes In
A miner, or in practice the pool that coordinates many miners, builds a block template: the list of transactions its machines will try to turn into the next block. To collect the most fees that fit, it fills the template with the highest fee rates first, working down until the block is full.
Linked transactions are handled as groups. If a transaction spends an output of another unconfirmed one, the parent must be in the block too, so the pair is ranked by their combined fee rate. That is why a high-paying child can pull a stuck parent into a block, the trick called child-pays-for-parent (CPFP). Bitcoin Core 31.0 (April 2026) rebuilt its mempool around these groups, a design called “cluster mempool”.
Within the rules, each miner sets its own policy. On 6 November 2026 most blocks went down to transactions paying about 0.12 sat/vB, but block 974,791, mined by CloverPool, took nothing below 1 sat/vB and left most of its space empty. Some pools also sell priority outside the blockchain, through so-called accelerator services; those payments never appear on-chain, so fee figures read from the blockchain leave them out. Whatever the miner picks, every node checks that the coinbase transaction claims no more than the new coins plus the block’s fees.
Choosing Your Own Fee
- Look at the queue before you send. mempool.space, or the same software on your own node, shows the next few blocks as they would be mined now. Bid above the bottom of the first one to be in the next block, or lower if you can wait. In early November 2026, 1 sat/vB was usually enough for the next block; Bitcoin Core’s own wallet won’t go lower unless you change its settings.
- Keep a way to raise the fee. If the queue fills after you send, replace the transaction with a higher-paying version (replace-by-fee, RBF) or, as the receiver, use CPFP. The guide to unsticking a transaction shows both.
- Make your transactions smaller. Fewer inputs cost less. Combining many small coins into one while fees are low saves fees later, but it publicly links those coins; the coin control lesson explains the trade-off.
Who Pays the Miners Today
Miners earn two things per block: the block subsidy, the new bitcoin each block creates (3.125 BTC since April 2024), and the fees inside it. Together they pay for the machines and electricity that make rewriting Bitcoin’s history expensive (see how proof of work secures Bitcoin). The total paid to miners, usually counted per year, is called the security budget.
In the 1,008 blocks up to block 974,800 (about a week, ending 6 November 2026), miners earned 3,169.93 BTC: 3,150 BTC of subsidy and 19.93 BTC of fees. Fees were about 0.63% of the total, about 0.02 BTC per block. The share moves a little from week to week; every week of 2026 so far has stayed between about 0.4% and 0.9%. Over a year that comes to roughly 164,250 BTC of new coins and about 1,000 BTC of fees, around $12.6 billion at about $76,000 per bitcoin. So today new issuance pays for almost all of Bitcoin’s security: each year’s new coins add about 0.8% to the roughly 20.1 million in existence, slightly diluting every holder, while people sending transactions pay very little. To check the split yourself, open mempool.space’s block fees versus subsidy chart and look at a year or more, not single days.
It hasn’t always been this lopsided. Yearly averages from mempool.space’s block data:
| Year | Fees per block | In dollars at the time | Share of miner revenue |
|---|---|---|---|
| 2017 | 1.84 BTC | about $10,600 | 12.8% |
| 2018 | 0.45 BTC | about $4,300 | 3.5% |
| 2019 | 0.37 BTC | about $3,000 | 2.9% |
| 2020 | 0.52 BTC | about $6,900 | 5.7% |
| 2021 | 0.42 BTC | about $20,200 | 6.3% |
| 2022 | 0.10 BTC | about $2,700 | 1.6% |
| 2023 | 0.44 BTC | about $14,800 | 6.5% |
| 2024 | 0.29 BTC | about $17,500 | 6.6% |
| 2025 | 0.033 BTC | about $3,300 | 1.0% |
| 2026 (to 6 November) | 0.020 BTC | about $1,500 | 0.6% |
Fees jump when many people want block space at once. On 20 April 2024, the day of the last halving, a token protocol called Runes launched in the halving block itself: block 840,000 carried 37.6 BTC in fees, and that day fees made up about three-quarters of everything miners earned. Then demand fell back. Counted in bitcoin, 2025 and 2026 have the lowest average fees per block of any year since 2010.
What Each Halving Does to the Split
Every 210,000 blocks the subsidy halves (see the halving lesson); the next halving, at block 1,050,000, is expected around early-to-mid 2028. Fees follow no schedule. Here is what the coming halvings would mean if fees stayed at today’s 0.02 BTC per block:
| Era | Subsidy per block | Fee share if fees stay at 0.02 BTC | Fees needed to pay miners today’s 3.14 BTC per block |
|---|---|---|---|
| Today (since April 2024) | 3.125 BTC | 0.6% | 0.02 BTC (today’s level) |
| 5th halving, around 2028 | 1.5625 BTC | 1.2% | 1.58 BTC, about 80 times today |
| 6th halving, around 2032 | 0.78125 BTC | 2.5% | 2.36 BTC, about 120 times |
| 7th halving, around 2036 | 0.390625 BTC | 4.8% | 2.75 BTC, about 140 times |
| 8th halving, around 2040 | 0.1953125 BTC | 9.2% | 2.95 BTC, about 150 times |
| Around 2140 | 0 | 100% | 3.14 BTC, about 160 times |
Two things stand out. The fee share roughly doubles at each halving without fees growing at all, so a rising share on its own says little. And paying miners the same number of bitcoin after 2028 would take about 1.58 BTC of fees per block: spread over a full block of about 1 million vB, an average of about 158 sat/vB in every block, day and night. In early November 2026, fees averaged about 2 sat/vB across each block (a few high payers pull the average up), and the median transaction in most blocks paid about 0.12 sat/vB.
Miners pay their bills in dollars, euros and other currencies, though, so price is the other lever (the profitability lesson shows how price sets a miner’s income). If bitcoin’s price doubles between halvings, the subsidy pays miners the same in dollars with no help from fees. Counted in coins, the budget shrinks on schedule: new coins are about 0.8% of the supply per year today, and will be about 0.4% after 2028, 0.2% after 2032 and 0.1% after 2036.
Why the Long-Term Budget Is an Open Question
- Two unknowns. Miners’ future income depends on bitcoin’s price and on demand for block space decades from now. Nobody can forecast either.
- Nobody knows how much security is enough. The economist Eric Budish argued in a 2018 paper that what Bitcoin pays miners, over and over, must stay large compared with what an attacker could gain from one attack, so the more value Bitcoin carries, the more it needs. Others answer that the right level isn’t fixed (see below).
- Fees are lumpy; the subsidy is steady. Princeton researchers showed in 2016 that when miners live mostly on fees, a block that happens to carry unusually high fees can be worth more to replace than to build on: a rival miner can mine a competing copy of it and, if that copy rather than the original ends up in the chain, keep its fees. This is called fee sniping. Bitcoin Core’s wallet has guarded against it by default since version 0.12 (2016), and several other wallets do the same. They usually set each transaction’s lock time (nLockTime) to the current block height, so it can go into the next block or a later one but never into a competing copy of the latest. A comment in Bitcoin Core’s code, written in 2015, notes that fee sniping isn’t a problem yet because the subsidy is high.
The Main Positions
Here are the main views, as their supporters would put them.
Demand for Block Space Will Grow Into the Role
Block space is fixed while use can grow. Systems built on top of Bitcoin, such as the Lightning Network, still open and close channels on-chain, and wider use means more demand for that settlement. In 2017 fees averaged 1.84 BTC per block for the whole year, more than the 1.58 BTC needed after 2028. And the subsidy fades slowly, leaving decades for the fee market to develop.
The Price Buys Time
Over Bitcoin’s history the subsidy has grown enormously in dollars even as it halved in coins, a point the Bitcoin Core contributor Anthony Towns made with data in a June 2026 developer discussion. As long as the price at least doubles from one halving to the next, miners’ income in dollars holds up without help from fees. After each of the first three halvings (2012, 2016 and 2020), the price more than doubled before the next one. Since the 2024 halving it came close, at its October 2025 peak, then fell back: each block’s subsidy was worth about $400,000 just before that halving and about $238,000 in November 2026.
Today’s Security May Be More Than Needed
Greg Maxwell, a long-time Bitcoin developer, argued in July 2026 that the amount of security is not a fixed requirement: users can adapt, for example by waiting for more confirmations, and Bitcoin may be heavily over-secured today compared with systems that run on a small fraction of its budget. On this view, a smaller budget later need not be a crisis.
The Trend Is a Warning
Others read the data the opposite way: fees were about 6.5% of miner revenue in 2023 and 2024, then about 1% in 2025 and 0.6% so far in 2026. If the budget must grow with the value Bitcoin carries, a shrinking subsidy and falling fees point the wrong way. Some also note that people sending transactions pay the fees, while holders, who also benefit from security, pay only when they move their coins.
A Permanent Small Subsidy: Tail Emission
The most debated fix is tail emission: stopping the halvings at some point and creating a small, fixed amount of new bitcoin in every block forever. The Bitcoin developer Peter Todd argued in 2022 that this would not keep inflating the supply, because a share of all coins is lost every year and the total would settle where new coins equal lost ones. In June 2026 a proposal on Delving Bitcoin, a forum where developers discuss protocol ideas, suggested 0.25 BTC per block from about 2040, combined with destroying part of every fee, so that the supply would grow in quiet years and shrink in busy ones. Supporters argue that a small, predictable payment is the only source of security that doesn’t depend on guessing future demand. It is controversial for three reasons:
- It breaks the 21 million limit, which many holders see as Bitcoin’s defining promise. Pieter Wuille, a long-time Bitcoin Core developer, called the 2026 proposal “a complete non-starter”, arguing that what sets Bitcoin apart is confidence that its properties are reliable and beyond anyone’s influence, and that such ideas should be tried in a separate currency instead.
- It would be a hard fork. Nodes running today’s rules reject blocks that create extra coins, so anyone who didn’t switch software would stay on the original chain, and the two would become separate coins (see who enforces Bitcoin’s rules).
- Someone has to pick the rate, with no way to know the right level, and new coins benefit whoever receives them at the expense of other holders, an objection Greg Maxwell raised.
Other ideas leave the 21 million limit alone, such as the developer Jameson Lopp’s suggestion to let the block size limit adjust to demand, shrinking when fees are low so that space stays scarce, or paying miners from outside Bitcoin through merged mining, where the same work also secures another system. Each has trade-offs, and the 2026 discussion ended without agreement on any of them.
Key Takeaways
- A fee is a bid for block space, ranked by fee rate (sat/vB); the amount sent plays no part.
- Full blocks don’t mean high fees: the price climbs only when more than a block’s worth of transactions bid well above the minimum.
- Fees were about 0.63% of miner revenue in the week to 6 November 2026. After the 2028 halving, holding miners’ income in bitcoin takes about 80 times today’s fees; with fees unchanged, holding it in dollars takes a doubling of the price.
- Whether fees can carry Bitcoin’s security is an open question with serious arguments on every side.
Check Your Understanding
1. Two transactions are waiting. A moves 2 BTC, takes 141 vB and pays 1,410 sats. B moves 0.005 BTC, takes 753 vB and pays 3,765 sats. Which one does a miner prefer, and why?
Answer
A. It pays 10 sat/vB, while B pays 5 sat/vB. B pays more in total but uses more than five times the space, and space is what the miner gives up. The amounts moved don’t enter into it.
2. Suppose the fee share of miner revenue goes from about 0.6% today to about 1.2% after the 2028 halving. Has the fee market grown? What would keep miners’ income where it is today?
Answer
Not necessarily: if fees stay at about 0.02 BTC per block, the share doubles only because the subsidy halved. Paying miners today’s roughly 3.14 BTC per block would take about 1.58 BTC of fees per block, about 80 times today’s level. A doubling of bitcoin’s price would instead keep their income the same in dollars.
3. Suppose part of the network adopts software with a tail emission of 0.25 BTC per block. What happens, and why is the idea controversial?
Answer
Nodes still running today’s rules reject every block that creates the extra coins, so the network splits into two chains, each its own coin: a hard fork. It is controversial because it breaks the 21 million limit, because someone has to choose the rate with no way to know the right level, and because new coins benefit their recipients at the expense of other holders. Supporters answer that it is the only source of security that doesn’t depend on future demand for block space.