The mempool holds tens of thousands of transactions and almost none of them compete with you. Bitcoin Core 29.1 lowered the minimum relay fee to 0.1 sat/vB in September 2025, and a permanent queue formed at that floor. It looks like congestion on a chart and behaves nothing like it.
Open any mempool visualiser and you see an alarming picture. Tens of thousands of unconfirmed transactions. Dozens of blocks worth of backlog. Then look at the recommended fee and it says 1 sat/vB, which is close to nothing.
People assume one of the two numbers must be wrong. Neither is. They describe different things, and the gap between them is the single most useful fact about Bitcoin fees right now.
What we measured today
These figures come from the live mempool and are re-measured every morning. They are not copied from anywhere.
The third and fourth numbers are the story. Roughly 69 percent of that entire backlog sits in a band between 0.1 and 0.2 sat/vB. The highest paying transaction anywhere in the mempool is 6.04 sat/vB, which in a genuinely congested market would be an unremarkable middle rate.
And the whole queue, all 85,351 transactions together, carries under a tenth of a bitcoin in fees. A single block's subsidy is 3.125 BTC. The entire waiting room is worth about three percent of one block's base reward.
What changed in September 2025
Bitcoin Core 29.1 lowered the default minrelaytxfee from 1 sat/vB to 0.1 sat/vB, a ninety percent reduction. The incremental relay fee moved with it.
Why the developers did it
The old floor had become a problem rather than a protection. Blocks were routinely including transactions below 1 sat/vB through direct submission and mining pool arrangements, and nodes that had rejected those transactions could not validate the resulting blocks as quickly, because they had to fetch transactions they had never seen. Lowering the floor brought node policy back in line with what miners were actually doing.
What it caused
Wallets exposing an economy or slow option started producing transactions at rates close to the new floor. Those transactions relay fine and enter mempools normally. They simply never get mined, because there is always something paying more.
Why the floor queue never drains
This is the part that makes it different from ordinary congestion, and it is worth being precise about.
In a genuine backlog, a fee spike is temporary. Demand exceeds capacity, fees rise, the marginal users postpone or pay more, and the queue clears within hours or days. The market resolves it.
The floor queue does not behave that way, for two reasons.
- The supply of floor rate transactions is effectively unlimited. Consolidations, dust sweeps and automated activity are perfectly happy to wait indefinitely, so the band refills as fast as it empties.
- The transactions in it are not price sensitive. Whoever sent them chose the cheapest option deliberately. There is no bidding war because nobody in that band wants to bid.
The result is a standing pool that looks like a queue and functions like sediment. It is not waiting for capacity. It is parked.
The number to remember
Only 2.12 percent of the mempool pays 1 sat/vB or more. If your transaction pays that, you are ahead of about ninety eight percent of what looks like a queue, and the backlog behind you is irrelevant to your confirmation time.
The distribution, band by band
Averages hide what is happening here. The shape of the distribution is the whole story, so here it is broken into bands.
| Fee band | Share of mempool | Behaviour | Will it confirm |
|---|---|---|---|
| 0.1 to 0.2 sat/vB | about 69 percent | Parked at the relay floor, refills constantly | Rarely, then dropped |
| 0.2 to 0.3 sat/vB | roughly 15 percent | Economy tier from wallet defaults | Only in quiet stretches |
| 0.3 to 1 sat/vB | roughly 14 percent | Below the clearing rate but close | Hours to days |
| 1 sat/vB and above | about 2.12 percent | The only genuine competition | Next block or two |
Two thirds of the visible backlog lives in the top row, and the top row is not going anywhere. That is why the total is such a poor guide to your wait.
| Version | Date | Default relay floor | Effect |
|---|---|---|---|
| Core 28 and earlier | to Sept 2025 | 1 sat/vB | Nothing cheaper was relayed at all |
| Core 29.1 | Sept 2025 | 0.1 sat/vB | Floor queue begins forming |
| Core 30 | Oct 2025 | 0.1 sat/vB | Unchanged floor, larger OP_RETURN limit |
| Core 31 | Apr 2026 | 0.1 sat/vB | Cluster mempool replaces ancestor limits |
How to read a mempool chart properly
Most visualisers colour by fee rate and stack by volume, which makes the total height dominate the picture. That height is the misleading part.
- Ignore the total. It tells you almost nothing about your wait.
- Find the next block projection. The lowest rate in the first projected block is the number that decides whether you confirm soon.
- Compare your rate to that. Above it means next block or close to it. Below means you are waiting on the shape of the distribution rather than on time.
- Check the top rate. If the highest bid anywhere is 6 sat/vB, there is no competition worth worrying about.
What this means for your transaction
If you paid 1 sat/vB or more, you are fine and the backlog is not your problem. Wait for the next block.
If you paid something starting with a zero, you are in the sediment, and waiting is not a strategy because the layer above you never empties. That is a fee bump, not a patience problem, and it costs a few hundred satoshis.
Why accelerator marketing loves this chart
A screenshot of a 44 block backlog is a persuasive sales asset. It is accurate, and the conclusion it invites is false.
When we measured actual paid accelerations, the median transaction being accelerated was already paying about 2.2 sat/vB, which was comfortably above the clearing rate. Those buyers were looking at a queue that had nothing to do with them.
What would change it
The fee market returns when block space is genuinely contested again. Historically that has meant a new form of on-chain demand: the 2017 retail wave, the 2023 and 2024 inscription and Runes activity. Something has to want block space badly enough to bid.
Until then the pattern holds. A large visible backlog, a low clearing rate, and a gap between them that traps anybody who chose the cheapest fee option in their wallet.
The long term question
Miner income comes from the block subsidy plus fees, and the subsidy halves every four years. A fee market this thin is comfortable for users and a genuine open question for mining economics. That debate is well beyond this page, but it is the reason developers watch these numbers closely.
The bottom line
The mempool is full of transactions that are not competing with you. The chart is real, the congestion is not, and the difference is a policy change from September 2025 that most fee advice has not absorbed.
Check your rate against the clearing rate rather than against the size of the queue. That single comparison answers the question the chart cannot.
See where you sit in the real queue
We show your effective rate, the current clearing rate, and how much of the backlog is genuinely above you. Free.
Frequently asked questions
Why is the Bitcoin mempool full when fees are so low?
Because most of what is in it is not competing for the next block. Bitcoin Core 29.1 lowered the minimum relay fee to 0.1 sat/vB in September 2025, and a permanent queue formed at that floor. Roughly two thirds of the mempool now sits between 0.1 and 0.2 sat/vB, which looks like a backlog on a chart but never clears and never bids against you.
What is the minimum relay fee in Bitcoin?
Since Bitcoin Core 29.1 in September 2025 the default minrelaytxfee is 0.1 sat/vB, down from 1 sat/vB. It is the cheapest rate at which a node will accept and relay a transaction. It is a relay policy, not a consensus rule, so individual node operators can configure it differently.
Does a big mempool backlog mean my transaction will be slow?
Not by itself. What matters is how much of the backlog pays more than you do. When only about two percent of the mempool pays 1 sat/vB or more, a transaction at that rate is ahead of nearly everything waiting and the total size of the queue is irrelevant to its confirmation time.
Why do transactions at the relay floor never confirm?
Because the supply of floor rate transactions is effectively unlimited and the senders are not price sensitive. Consolidations and automated activity are content to wait indefinitely, so the band refills as fast as it drains. Unlike a genuine fee spike, there is no market pressure to resolve it.
How much in fees is the whole Bitcoin mempool worth?
Under a tenth of a bitcoin across tens of thousands of transactions, as measured on this page and re-measured daily. For comparison a single block currently pays a 3.125 BTC subsidy, so the entire backlog is worth about three percent of one block's base reward.
Will Bitcoin fees go back up?
They rise when block space is genuinely contested again, which historically has required a new source of on-chain demand such as the 2017 retail wave or the inscription and Runes activity of 2023 and 2024. Until something wants block space badly enough to bid, the current pattern of a large backlog and a low clearing rate holds.