Every Ethereum block is a small auction, and most people trading on-chain never know they are bidding in it. When you send a swap, it does not go straight into a block. It waits in a public queue where anyone can read it, copy it, and decide whether there is money to be made by placing their own transactions around yours. That profit has a name. It is called MEV, short for maximal extractable value, and it has quietly become one of the largest recurring revenue streams in crypto.
The number is not small. Researchers who track on-chain extraction estimate that MEV has pulled well over $1 billion out of Ethereum users since the practice was first measured. Some estimates put the lifetime figure higher. The exact total is disputed, because a lot of it hides inside ordinary-looking trades, but the direction is not.
MEV is the profit hidden in the order of transactions
The Ethereum documentation defines it precisely. MEV is "the maximum value that can be extracted from block production in excess of the standard block reward and gas fees by including, excluding, and changing the order of transactions in a block."
Read that again, because the whole thing lives in one phrase: changing the order. Whoever decides the sequence of transactions inside a block holds a lever. Move a trade up, and it executes first. Push it down, and it fills at a worse price. Drop it entirely, and it never happens. On a public blockchain where pending transactions sit in the open, that power is worth a great deal, and an entire industry has grown up to capture it.
None of this requires breaking the rules. MEV is extracted using the same transaction system everyone else uses. That is what makes it hard to stamp out.
It used to be called miner extractable value
The M originally stood for miner. Under proof-of-work, miners chose which transactions went into a block and in what order, so the profit from that ordering flowed to them. After Ethereum switched to proof-of-stake in September 2022, the job of proposing blocks moved to validators, and the term was broadened to maximal extractable value to cover the wider set of players who now compete for it. If the difference between how the two systems agree on blocks is still fuzzy for you, our explainer on how proof-of-work and proof-of-stake reach consensus lays out the mechanics.
The rebrand mattered because extraction is no longer done by the block proposer alone. It is a supply chain now, and each link takes a cut.
Four ways value gets pulled out of a block
Most MEV falls into a handful of categories. Some of it is arguably useful. Some of it is a straight tax on the person trading.
| Type | What the bot does | Who pays |
|---|---|---|
| DEX arbitrage | Buys a token cheap on one exchange and sells it dearer on another in the same transaction | Nobody directly, it corrects a price gap |
| Liquidations | Races to close an under-collateralised loan and claim the liquidation fee | The borrower being liquidated |
| Sandwich attack | Buys just before your trade and sells just after, pocketing the price it moved | You, the trader |
| Backrunning | Places a transaction immediately after a known event, such as a new listing, to catch the reaction | Varies |
Arbitrage is the benign end. When one venue prices ether at $4,000 and another at $4,010, a bot buying on the cheap side and selling on the expensive side nudges both back toward the same number. That is a service, in a rough way. The same logic drives bots that watch lending markets and pounce the instant a position falls under water. If you have ever wondered how a loan gets closed the second it dips below its threshold, the answer is usually an MEV searcher, and our piece on how margin liquidations work covers the borrower's side of that trade.
The sandwich is the ugly end. It exists only to take money from a trader who was going to trade anyway.
How a sandwich attack actually works
Picture a large buy order sitting in the public queue. A bot sees it and moves in three quick steps.
- It submits its own buy first, paying extra gas so its transaction jumps the line and lands ahead of yours.
- Your trade then executes at a slightly worse price, because the bot's purchase already pushed the market up.
- The bot immediately sells into that inflated price, and the gap between what it paid and what it sold for is the profit.
Your slippage tolerance, the amount of price movement you told your wallet to accept, becomes the bot's margin. Set it too loose and you hand the bot more room to squeeze. This is closely tied to how automated market makers price trades against a pool, the same mechanism behind impermanent loss for liquidity providers.
The most infamous practitioner is a bot tied to the address jaredfromsubway.eth. On-chain trackers credit it with roughly 117,000 ETH, worth around $295 million, harvested through sandwich trades since it appeared in March 2023. It has sandwiched high-profile users, including a token swap by Vitalik Buterin in May 2026. The bot is not invincible. In June 2026 an attacker used dozens of fake token contracts to drain about $7.5 million from it, a rare case of the predator getting eaten.
Sandwich activity has thinned out lately. Monthly extraction from sandwiches fell from around $10 million in late 2024 to roughly $2.5 million by October 2025, a decline of about three quarters, as more trades started routing through channels the bots cannot see.
Who takes the cut: searchers, builders, relays, validators
Extraction is a pipeline with four roles.
Searchers run the algorithms. They scan the mempool for opportunities, build the transactions that capture them, and bid high gas to get included. Builders assemble those transactions into a full, profit-optimised block. Relays sit in the middle, collect blocks from competing builders, and pass the most valuable one along. Validators, the proposers, run software that accepts the winning block and publishes it to the network. Everyone in that chain earns something.
This design has a name: proposer-builder separation. It was introduced to stop validators from needing to run their own extraction operations, which would have pushed staking toward a few sophisticated giants. The tool that made it practical is MEV-Boost, middleware that lets an ordinary validator plug into the builder market and collect the extra income without doing the work. Flashbots, which built it, estimates the software can lift a validator's staking rewards by more than 60%.
Adoption is close to total. More than 90% of Ethereum validators now run MEV-Boost. That solved the centralisation worry at the validator layer, but it moved the concentration somewhere else. Three relays route roughly 85% to 88% of all MEV-Boost blocks, and a single builder, Titan, assembles better than half of them on its own. The lever did not disappear. It just changed hands.
Can you avoid paying it
Mostly, yes, if you take a few steps. Tightening your slippage setting shrinks the room a sandwich bot has to work in, though set it too tight and legitimate trades start failing. Trading through a private transaction service, sometimes called a protected RPC, keeps your order out of the public queue so bots never see it coming. Many wallets and exchanges now route swaps this way by default. Larger orders can also be broken into smaller pieces so no single trade is fat enough to be worth sandwiching.
Frequently asked questions about MEV
Is all MEV harmful?
No. Arbitrage and liquidations keep prices accurate and lending markets solvent, and those functions have to happen regardless of who profits from them. The predatory slice, mainly sandwiching, is the part that draws the complaints, and it is also the part protective tools target first.
Does MEV exist on other chains?
Yes. Any blockchain where transactions wait in a public queue before confirmation has the same opening, so Solana, BNB Chain, and others see their own versions. Ethereum is simply where it has been measured longest and in the most detail.
Is MEV the same as a gas fee?
No. A gas fee is the posted price you pay the network to process your transaction. MEV is the extra value someone else captures by controlling where your transaction lands, and you often pay it without any line item showing you did.
How would I know if I got sandwiched?
You usually will not, unless you go looking. The trade still confirms and your tokens still arrive, so nothing appears broken. The tell is in the fill price. If a swap settled noticeably worse than the quote you were shown, and a block explorer shows a buy of the same token landing just before yours and a sell landing just after from the same address, that is the signature of a sandwich. Wallets with built-in protection will sometimes flag it, but many traders simply absorb the cost without ever checking. Wide slippage settings on volatile tokens are where most of it happens.
Why it matters for anyone trading on-chain
MEV is not a bug that a future upgrade will delete. It is a structural feature of open, ordered blockchains, and the same transparency that lets you verify every transaction is what lets a searcher read yours before it settles. The research effort of the past few years has been less about ending extraction than about making it visible, competitive, and shared more evenly, so that the profit does not quietly pool with whoever happens to control block order. That is why work on MEV now reads less like a war on bots and more like plumbing, rerouting where the value flows rather than pretending it can be switched off. Even a perfectly fair version of MEV would still be a cost of trading on an open ledger.
For a trader, the practical takeaway is smaller and more useful. The price you see on screen is not always the price you get, and the difference is often someone else's business model. Knowing that changes how you set slippage, which routes you trust, and how large an order you are willing to send into the open.