The Bitcoin Mempool: Where Your Transaction Waits in Line

On April 19, 2024, the miner who found block 840,000 — the halving block — earned 3.125 BTC in new coins and roughly 37.6 BTC in transaction fees. For one block, the fees were twelve times the subsidy. Thousands of people were bidding frantically to get into that specific block, and the price of a spot in it went vertical. That auction happened in a place most bitcoin users have never looked at directly: the mempool.

If you have ever sent bitcoin and watched it sit unconfirmed for two hours while your wallet said "pending," you have already met the mempool. Understanding it is the difference between overpaying every time and knowing exactly when to press send.

There is no single mempool

The mempool — short for "memory pool" — is the waiting room for unconfirmed transactions. When you broadcast a transaction, your wallet hands it to a node, that node checks it against consensus and policy rules, and if it passes, the node holds it in RAM and relays it to its peers. Within a few seconds it has propagated across most of the network.

Critically, every node keeps its own mempool. There is no canonical list. Bitcoin Core defaults to holding 300 MB of pending transactions; when that fills, the node evicts the lowest-fee ones and raises its minimum relay fee. A node that just restarted has an empty mempool. A node with a smaller memory limit is holding a different set than yours. Sites like mempool.space show you one node's view, not the network's — a useful approximation, not gospel.

This is not a flaw. The mempool is not consensus. Nothing in it is final, nothing in it is guaranteed, and the only thing that has ever counted as "settled" in Bitcoin is a transaction buried under proof-of-work.

The block space auction

Every block has a hard ceiling of 4 million weight units, which works out to roughly 1 to 2 MB of real transaction data depending on what kind of transactions are in it — usually somewhere between 2,000 and 4,000 transactions. That supply is fixed and arrives, on average, every ten minutes. Demand is not fixed. That mismatch is the entire fee market.

Miners are profit-maximizing, so they build blocks by sorting pending transactions by fee rate — satoshis per virtual byte, written sat/vB — and taking the most valuable ones first. Note that this is fee rate, not total fee. A 200-vB transaction paying 2,000 sats (10 sat/vB) beats a 2,000-vB transaction paying 10,000 sats (5 sat/vB), even though the second one pays five times more in absolute terms. Miners are selling space, not processing payments.

This is why the size of your transaction matters as much as the amount you are sending. Sending 0.5 BTC costs the same as sending 5,000 BTC if both use one input and two outputs. But a wallet stuffed with 40 tiny UTXOs from months of small buys will produce a fat transaction that costs a fortune to move at any fee rate. Consolidate during quiet periods, not during a rush.

What actually drives congestion

Fee spikes are almost never random. Historically they cluster around a few causes:

  • Speculative mania. In December 2017, at the top of that cycle, average fees crossed $50 and the mempool held hundreds of thousands of transactions for days.
  • New on-chain use cases. The 2023 Ordinals and BRC-20 wave, and the Runes launch that coincided with the 2024 halving, both put enormous non-payment demand on block space in a short window.
  • Block variance. Blocks arrive on a Poisson distribution. A 45-minute gap between blocks — which happens regularly — backs up the queue on its own, then clears when two blocks land three minutes apart.

Conversely, weekend evenings in a quiet market are frequently clear enough that 1–2 sat/vB confirms within a few blocks. Between the extremes, most of 2025 and 2026 has hovered in a low single-digit to low-double-digit sat/vB range.

If you get it wrong: RBF and CPFP

Underpaying is recoverable. Bitcoin gives you two escape hatches.

Replace-by-fee (RBF), specified in BIP 125, lets you rebroadcast the same transaction with a higher fee. Nodes replace the old version in their mempool with the new one. Most modern wallets expose this as a "bump fee" or "speed up" button. Signal it when you send, and a stuck transaction becomes a thirty-second fix.

Child-pays-for-parent (CPFP) works from the receiving side. If someone sent you coins at too low a fee rate, you can spend that unconfirmed output in a new, high-fee transaction. Miners evaluate the pair as a package, so the child's fee drags the parent into a block. Useful when you did not send the transaction and cannot replace it.

The one thing that does not work is waiting forever. If your transaction falls below every node's minimum relay fee, it gets evicted and effectively disappears — the coins were never spent, and after your nodes purge it you can simply re-send at a sane rate.

What to actually do

Three habits cover most of it. Check before you send. Open a mempool explorer, look at the projected next few blocks, and pick the tier that matches your actual urgency — a cold-storage deposit does not need to confirm in ten minutes. Turn on RBF by default so a bad guess is never a stuck guess. Batch and consolidate deliberately. If you buy sats weekly, sweep them into a single UTXO on a cheap Sunday rather than discovering the cost during the next fee spike.

And understand why this matters long-term. The block subsidy halves every four years and hits zero around 2140. Fees are what will pay for Bitcoin's security after that. The mempool is not a bug in a system that should be free — it is the market that has to eventually fund the whole thing.

If you have read this far, you already understand Bitcoin better than most people who own it. BitCloset makes heavyweight apparel for people who chose it on purpose.

Further reading

Back to blog

Leave a comment