Bitcoin Network Fees Explained: How They Work and How to Pay Less

Two people send bitcoin on the same afternoon. One moves fifty dollars, the other moves fifty million, and both pay the same fee.

A Bitcoin network fee is the amount you pay the miner who confirms your transaction, priced by how much block space the transaction takes up rather than how much bitcoin you are sending. The size of the payment does not enter into it, which is why those two senders are charged the same.

This guide covers what the BTC network fee actually pays for, the arithmetic behind the number your wallet shows you, the three things that make a transaction expensive, and the practical moves that reduce what you pay.

Key Takeaways

• A Bitcoin network fee is paid to the miner who confirms your transaction, not to any company, exchange or wallet provider.
• The fee is calculated as transaction size in virtual bytes multiplied by a fee rate in sat/vB. The amount of bitcoin you send does not appear in that formula.
• What makes a transaction expensive is the number of inputs it spends, the number of outputs it creates, and the address format involved.
• Native SegWit (bc1q) is the cheapest address type for an ordinary one-payment send. Taproot (bc1p) becomes cheaper once a transaction spends several inputs at once.
• Bitcoin Core 30.0, released in October 2025, cut the default minimum relay feerate from 1 sat/vB to 0.1 sat/vB, so the old "1 sat/vB floor" no longer describes how nodes behave.
• Fees have ranged from fractions of a cent to a record daily average in April 2024. Conditions have been quiet through most of 2026, but they can change within hours.
• Every fee figure has a shelf life. Check current rates before you send rather than trusting a number you read weeks ago.

What Are Bitcoin Network Fees?

A Bitcoin network fee, also called a bitcoin transaction fee or bitcoin miner fee, is a payment to the miner who includes your transaction in a block. It is not a charge levied by Bitcoin itself, and no company collects it.

The money goes to whichever mining operation finds the next valid block. Miners collect all the fees from the transactions they include, add them to the block subsidy of newly issued bitcoin, and pay the total to themselves in a special first transaction in the block called the coinbase transaction.

Your wallet never creates an output labelled "fee" the way a bank statement lists a charge. A Bitcoin transaction lists what you are spending on one side and where it is going on the other, and the fee is whatever is left over. Spend 100,000 satoshis, send 90,000 to a friend and 9,800 back to yourself as change, and the missing 200 satoshis are the fee. The miner claims the gap.

Fees exist for two reasons, and both were in the design from the start. Satoshi Nakamoto's whitepaper describes transaction fees as the incentive that keeps miners working once new coin issuance tapers off. The second reason is more practical. Block space is limited, and because occupying it costs money, flooding the network with junk transactions is expensive rather than free.

How Bitcoin Transaction Fees Are Calculated

There is one formula, and it has two inputs:

Total fee (in satoshis) = transaction size (in virtual bytes) × fee rate (in sat/vB)

Take the three terms in turn.

A satoshi is the smallest unit of bitcoin, one hundred-millionth of a BTC, or 0.00000001 BTC. Fees are quoted in satoshis because quoting them in whole bitcoin would produce a string of zeroes.

A virtual byte (vB) measures how much room your transaction occupies in a block. Blocks are capped at 4,000,000 weight units, and one virtual byte equals four weight units, which gives a practical ceiling of 1,000,000 vB per block. Weight units exist because Segregated Witness (SegWit) introduced a discount on signature data in 2017, and the network needed a way to measure size that reflected the discount while staying backward compatible.

The fee rate, written sat/vB and read as satoshis per virtual byte, is the price per unit of space that you choose to bid. Older articles quote fees in sat/byte. That unit predates SegWit and has not been the working measure for years, so treat any page still using it as out of date.

Put the formula to work. A standard transaction spending one input and creating two outputs, using a Native SegWit address, comes to roughly 141 vB:

To convert any of those into your own currency, multiply the satoshi figure by the current bitcoin price and divide by 100 million. You can see exactly what any confirmed transaction paid by looking it up on the Bitcoin.com block explorer, which shows both the total fee and the fee rate.

So when someone asks how much it costs to send bitcoin, the honest answer is that it depends on two numbers, only one of which is under your direct control at the moment you press send. You choose the rate. The size was largely decided earlier, by how your coins arrived in your wallet.

Why the Amount You Send Does Not Change the Fee

The value of a transaction plays no part in the calculation. A transfer worth ten dollars and a transfer worth ten million dollars, built the same way, cost the same to confirm.

The reason lies in how Bitcoin tracks ownership. There are no account balances. Instead the network keeps a list of discrete chunks of bitcoin called unspent transaction outputs, usually shortened to UTXOs. Your wallet balance is the sum of the UTXOs you control, in the same way that the cash in your pocket is the sum of the individual notes.

Spending works like spending cash. You hand over whole notes and receive change. If you hold a single 1 BTC UTXO and want to send 0.1 BTC, your wallet builds a transaction that spends the whole 1 BTC chunk, sends 0.1 BTC to the recipient and returns the remainder to you as a change output. The size of that transaction depends on how many notes you handed over and how many you got back, not on the number written on them.

The result can run against intuition. Someone moving a fortune held in one large UTXO pays very little, while someone moving pocket change scraped together from forty small UTXOs pays a lot, because their transaction is physically bigger.

What Makes a Transaction Big or Small

Three variables decide the size of your transaction.

• Inputs: Each UTXO you spend adds a chunk of data carrying the proof that you are allowed to spend it. Inputs are the most expensive component by a wide margin.
• Outputs: Each destination adds a smaller chunk. Most sends create two: the payment and your change.
• Address type: The format of the addresses involved changes how many bytes each input and output takes.

Component sizes, for a single-signature spend:

Add roughly 10.5 vB of overhead and you can size any ordinary transaction yourself. For the standard shape of one input and two outputs, the totals land like this:

Moving from a Legacy address to Native SegWit cuts that transaction by 38% for identical results.

Why Taproot Is Not Always Cheaper

Look again at the table. Taproot costs more than Native SegWit for a routine payment, which cuts against the common shorthand that a newer address format is always the cheaper one.

The arithmetic explains why. Taproot uses Schnorr signatures, so a P2TR input is the smallest of any type at 57.5 vB, around 15% below a Native SegWit input. Its outputs, though, are 43 bytes rather than 31, because a Taproot output carries a full 32-byte public key instead of a 20-byte hash. On a transaction with one input and two outputs, the larger outputs outweigh the smaller input. Bitcoin Optech documented this trade-off when the upgrade shipped.

Flip the ratio and Taproot pulls ahead decisively. Sweeping ten small UTXOs into one:

Taproot also wins on multisig and Lightning channel operations, where the savings are far larger than anything single-signature users see.

Dust, and Why Small Coins Can Become Unspendable

A UTXO worth less than the fee required to spend it is effectively stranded. The network calls very small outputs dust and refuses to relay transactions that create them, with the cut-off sitting somewhere around 300 to 550 satoshis depending on address type.

Dust is a slow-motion problem. Collect enough tiny UTXOs during a quiet period and a future fee spike can leave them costing more to move than they hold. The fix is to consolidate while rates are low, which is covered below.

Why Bitcoin Fees Rise and Fall

Block space is auctioned. Roughly every ten minutes a block arrives with room for about 1,000,000 vB, and everyone waiting bids for a place in it. Miners build each block by taking the highest fee rates first, because they are paid per unit of space rather than per transaction. A small transaction paying 40 sat/vB beats a large one paying a bigger total at 4 sat/vB.

When demand exceeds what the next few blocks can absorb, a queue forms and the price to skip it goes up. When demand falls away, the queue empties and the price collapses toward the relay minimum. No central authority sets the rate at any point, because it emerges from thousands of senders bidding against each other for the same limited space.

Two practical consequences follow. First, the fee your wallet suggests is a prediction of what miners will accept over the next few blocks, drawn from recent history and the current queue. Predictions miss. Second, timing matters, since demand tends to thin out at weekends and outside US and European business hours.

That queue is called the mempool. A transaction that lands at the back of it waits there, sometimes for days, and the sender can either raise the fee or sit it out.

What Bitcoin Network Fees Have Actually Cost

Fee history is more useful than any current snapshot, because it shows the range.

The April 2024 figures come from Coin Metrics data reported by Forbes and remain the high-water mark for USD-denominated fee revenue, though the longer series on Bitcoin Charts puts that single day in proportion against the years either side of it.

The 1 sat/vB Floor Is No Longer the Floor

For years, 1 sat/vB was treated as the hard minimum, because that was the default rate below which a Bitcoin node would refuse to pass a transaction along. That default changed.

According to the Bitcoin Core 30.0 release notes, published in October 2025, the default minimum relay feerate and the default incremental relay feerate were both reduced to 0.1 satoshis per vB, and the minimum block feerate dropped to 0.001 satoshis per vB. The change reflected reality: blocks had been confirming sub-1 sat/vB transactions for some time.

Two caveats matter, and the release notes are explicit about both. Relay depends on how widely the lower defaults are adopted across the network, so a very low-fee transaction is not guaranteed to propagate or confirm. And other minimum feerates, including the dust threshold and the rates your wallet uses, were left unchanged. In practice most wallets still floor you at 1 sat/vB. The point is that the protocol-level constraint people cite as fixed has moved, and articles still printing "1 sat/vB minimum" as a rule of Bitcoin are describing software from 2024.

How to Pay Less in Bitcoin Network Fees

Six moves, roughly in order of how much they save.

• Use a Native SegWit or Taproot address: The largest single saving available, and it costs nothing. A bc1q address cuts a routine send by 38% against a Legacy 1 address at any fee rate. Most modern wallets default to one of the newer formats, but coins received into an old address years ago still carry the old cost when you spend them.
• Consolidate small UTXOs when rates are low: Sweeping many small coins into one large one costs a lot in a busy period and very little in a quiet one. Doing it early means a future spike cannot strand them. Taproot addresses are worth using for this specifically.
• Batch payments into one transaction: Paying five people in five transactions means five sets of overhead and five inputs. One transaction with five outputs is substantially smaller than the sum of its parts.
• Match the confirmation target to real urgency: Most sends are not urgent. Choosing an economy tier when sending bitcoin between your own wallets, or on a payment nobody is waiting for, costs nothing but patience.
• Use the Lightning Network for small, frequent payments: Settling off-chain sidesteps the on-chain fee for everyday amounts, at the cost of committing funds to a payment channel first.
• Send with fee bumping available: Broadcasting in a way that lets you raise the fee afterwards means you can start low and increase only if conditions turn. Full replace-by-fee has been the default node policy since Bitcoin Core 28.0, as recorded in the Bitcoin Core mempool policy documentation, and the mechanics are covered on our mempool page.

The Fee You Are Quoted vs the Fee the Network Charges

A number worth knowing: the "network fee" a platform shows you at withdrawal is often not the fee that reaches the miner.

The mechanism is straightforward and not in itself improper. A custodial service, usually an exchange, estimates what the on-chain fee will be, charges you that estimate, then constructs and broadcasts the transaction itself, paying whatever the network actually requires. Several things can open a gap between the two figures:

• The quoted fee may be a flat rate per withdrawal rather than a live estimate, in which case it is right on average and wrong nearly every time.
• The estimate may carry a buffer so the platform is never out of pocket when rates move mid-processing.
• The withdrawal may be batched with other customers' withdrawals into a single transaction, which makes the real per-customer cost far lower than an individual send would be, while each customer is charged as though they sent alone.

None of that is concealed, and most platforms disclose it in their fee schedules, but you will not see the difference unless you go looking. To check, take the transaction ID from your withdrawal history and look it up on the Bitcoin.com block explorer. The explorer shows the fee that was genuinely paid, and comparing it against what you were charged tells you the rest.

This is one of the practical arguments for holding your own keys. With a self-custody wallet the fee you set is the fee the miner receives, with no intermediary in between.

What Happens to Fees as the Block Subsidy Shrinks

Miner income has two sources: the block subsidy of newly created bitcoin, and transaction fees. The subsidy halves roughly every four years. It fell to 3.125 BTC per block in April 2024, is scheduled to fall to 1.5625 BTC around April 2028, and reaches zero in the 2140s. Fees are what remain.

The awkward part is the current ratio. Through late 2025 and much of 2026, fees have contributed on the order of 1% of total miner revenue. The subsidy is carrying network security almost entirely, and it is scheduled to keep shrinking.

There is no settled answer to what happens next, and anyone who tells you otherwise is guessing. The two positions, fairly stated:

• Fees will be sufficient: Block space is permanently scarce, and demand for final settlement should grow as the network is used for larger and more valuable transfers. Layer-two systems batch enormous volumes of activity into comparatively few on-chain transactions, each of which competes for space. On this view a smaller number of high-value settlements can support a large security budget, and the 2023 to 2024 inscription episodes demonstrated that new sources of demand appear without warning.
• Fees may not be sufficient: Fee revenue has been volatile and has trended down since 2024. Security spending falls as the subsidy falls, and if fees do not rise to compensate, the cost of attacking the network drops with it. Critics of the optimistic case note that hoping for demand is not a plan.

The live version of this argument is about what block space should be used for. Bitcoin Core 30.0 raised the default data-carrier limit for OP_RETURN outputs from 83 bytes to 100,000 and allowed multiple such outputs per transaction, a change node operators can reverse in their own configuration. Supporters treat it as policy neutrality that reflects what was already happening. Opponents see a subsidy for non-monetary data at the expense of payments. Both camps agree it affects competition for the same space your transaction needs. Neither can impose an outcome, because Bitcoin has no mechanism for settling an argument like this by decree, and the policy that prevails is whichever one enough node operators choose to run.

Conclusion

A Bitcoin network fee is a payment to the miner who confirms your transaction, priced by the space it occupies and set by an open auction for that space. Size comes from the number of coins you spend, the number of destinations you create, and the address formats involved. Value never enters the calculation.

Fee conditions have been quiet for most of 2026, with rates near the low end of their historical range, though the record set in April 2024 is a reminder of how quickly that can reverse. The controllable parts are the ones worth acting on: use a modern address format, tidy up small coins while sending is cheap, and pay for speed only when speed is what you need. Once the transaction is broadcast the fee is set, and everything after that point happens in the mempool.