Bitcoin fee calculatorThe fee for the wait you want, read live from the mempool.

Fee rate is the only number miners look at. Every block has room for about one million virtual bytes, and miners fill it with the transactions paying the most per byte. A rate that clears the next block does not wait. A rate below it waits for everyone above it to be mined first.

Total fee is what leaves your wallet on top of the payment. At the rates the network has run for most of 2026 a normal payment costs a few hundred satoshis at most, which is cents. The dollar figure uses the live BTC price.

Queue ahead is the part almost no wallet shows you. We add up the size of every waiting transaction that pays a higher rate than yours, straight from the live fee histogram. That is the exact amount of block space that has to be mined before yours. Divide by one block and you get your place in line, shown as the amber cell on the ruler.

Typical wait is the time by which half of all outcomes have mined your block, given your place in line and how fast blocks have actually arrived over the last difficulty period. The second line is the time by which nine in ten have. The odds strip shows the same thing at four fixed points.

Size comes from three parts: a fixed overhead, the inputs you spend, and the outputs you pay to, one of which is usually your change. Each address type has its own costs because the witness discount introduced with SegWit counts signature data at a quarter of its raw size.

The overhead is 10.5 vB for SegWit and Taproot transactions and 10 vB for legacy ones. The calculator rounds the total up to a whole byte, the way wallets do.

Inputs are what make transactions expensive

An input costs two to four times what an output costs, because it carries the signature. A wallet that has received many small payments has to spend many inputs to send one large amount, and that is how people end up with a 700 vB transaction for an ordinary payment. Ten Native SegWit inputs and two outputs come to 753 vB, more than five times the size of a one input payment. At the same rate it costs five times as much.

Two habits keep this cheap. Consolidate small coins into one when the queue is quiet and the floor rate is enough, and prefer Native SegWit or Taproot receiving addresses. A legacy input costs more than twice a Taproot input every time it is spent.

Taproot is not always the smallest

Taproot inputs are the cheapest to spend, but Taproot outputs are 12 bytes bigger than Native SegWit ones. For a simple one input, two output payment the two land within 13 vB of each other. The difference that matters is legacy against anything newer.

Most people expect that paying a little more gets them a little sooner. The queue does not work like that. It has a cliff, and the cliff sits at whatever rate the next block is currently clearing.

Above the cliff the wait is the time to the next block, whatever you pay. Doubling the rate above it buys nothing at all. Below the cliff the wait depends on how much is queued between you and it, and in 2026 that is a very particular shape: a thin layer of transactions paying one sat/vB or more, then a thick slab sitting just above the relay floor. Since Bitcoin Core 29.1 lowered the default relay floor to 0.1 sat/vB in September 2025, that slab has been there almost every day, refilling as fast as blocks drain it.

So a rate a hair above the slab often clears within the hour, and a rate inside it can wait for days. The ruler in the result makes this visible: watch how the amber cell jumps when you switch from Economy to My rate and type a value a fraction lower.

The arrivals problem

The queue ahead of you is measured right now. Over the next hour new transactions arrive, and any that pay more than you join in front. For a rate near the top of the market that barely matters. For a rate near the floor it is most of the story, which is why the calculator calls a long estimate the best case. If timing matters, pick the next rung up. The difference is usually a handful of satoshis.

For the full measured picture of that slab, and why it looks like congestion when it is not, read why the mempool is full but fees are low.

One habit protects every option on that list: send with Replace By Fee available. Since Bitcoin Core 28, nodes treat every unconfirmed transaction as replaceable, so the real question is whether your wallet shows a bump button. If it does, a cheap rate is a safe gamble, because a stuck payment can be lifted later for the difference. Our fee bump calculator gives the exact amount, and the RBF guide shows where the button is in each wallet.

1. Using a fee guide written for 2017 or 2021. Rates of 20 or 50 sat/vB were normal then. In 2026 the median block has paid about 1 sat/vB for over a year. Old advice turns a ten cent payment into a ten dollar one.
2. Trusting a wallet's slow preset without looking at it. Some wallets set their slowest option at the relay floor. That is a rate inside the slab, and it can wait for days. Check the number, not the label.
3. Paying for priority above the next block. Once you clear the next block's cutoff, nothing pays you back for going higher. The cheapest way to be fast is the lowest rate that still lands in block one, which is what the Next block option shows.
4. Forgetting what the receiver waits for. Exchanges credit after their own number of confirmations, often three. The first one is the only one your fee controls; the others come one per block whatever you paid.
5. Spending dust. A coin worth less than the fee to spend it costs money to move. If your wallet has dozens of tiny inputs, consolidate in a quiet hour or leave them.

If you have already made the second mistake, it is fixable for a few hundred satoshis at most. What each hour of waiting means tells you when to act, and the bump calculator tells you how much.

Everything is computed in your browser from public data. Nothing you enter is sent to us.

• Rates come from mempool.space's precise fee estimates, refreshed every minute. Next block, about 30 minutes, about an hour, economy and the minimum relay rate are its five tiers.
• The queue is mempool.space's fee histogram: how many vbytes are waiting at each rate, which we add up above your rate. A block holds just under one million vbytes once the miner's own transaction is counted, so we use 997,000.
• Block timing uses the average interval of the current difficulty period instead of a flat ten minutes, so the odds reflect how fast blocks have really been coming.
• Sizes assume every input and output uses your address type. A payment to a different kind of address changes one output by up to 12 vB, a fraction of a satoshi at today's rates.

What it cannot know: transactions that will arrive after you, and whether a miner is about to find a block. The first is why long estimates are labelled best case. The second is why you see odds.