What Is MEV? How Sandwich Attacks Work

When you press swap on a decentralized exchange, your transaction is not executed the moment you sign it. It waits to be put into a block, and whoever orders that block decides what runs before and after it. The value that ordering power can capture has a name: maximal extractable value, or MEV.

Some MEV is useful, and some of it is taken straight out of your trade. This guide explains who extracts it, works through a sandwich attack on the real Uniswap formula so you can see exactly what it costs, compares how Ethereum, Base, Arbitrum, BNB Chain, Solana and Robinhood Chain differ, and is honest about what each protection does and does not cover.

What MEV is

ethereum.org defines it as “the maximum value that can be extracted from block production” beyond the standard block reward and gas fees, by including, excluding or reordering the transactions in a block. The term comes from a 2019 paper, Flash Boys 2.0, which called it “miner extractable value”; after Ethereum moved to proof of stake, “miner” became “maximal”.

The paper also described the first visible symptom: bots in priority gas auctions, “competitively bidding up transaction fees in order to obtain priority ordering,” and clogging the network for everyone else in the process.

Who takes it

• Searchers run software that scans pending transactions and chain state for profit, and submit their own transactions to capture it.
• Builders assemble those transactions, plus ordinary ones, into full blocks.
• Relays collect blocks from many builders and pass the most profitable one along.
• Validators (proposers) sign and publish the block. On Ethereum, validators can run MEV-Boost, which Flashbots describes as “open source middleware run by validators to access a competitive block-building market.”

Splitting the job between builders and proposers is called proposer-builder separation. In ethereum.org’s words, it “creates an auction market, where builders negotiate with validators selling blockspace.”

The good, the neutral and the bad

ethereum.org is blunt about the last one: sandwich trading results in “an unequivocally worse experience for users.”

How a sandwich works

A searcher watches for a large pending swap, then places two transactions around it:

1. Front-run: buy the same token just before the victim, pushing the price up.
2. Victim: the swap executes at the higher price and receives fewer tokens.
3. Back-run: sell what was bought in step 1 at the price the victim just pushed even higher.

What makes it possible is your slippage tolerance: the worst price you told the exchange you would accept. MEV Blocker’s documentation puts it directly: “Setting your slippage tolerance too high, however, leaves room for searchers to sandwich your trades.” The attacker’s job is to push the price to exactly your limit, and no further, so your transaction still goes through.

Worked example on the Uniswap v2 formula

A pool holds 1,000 ETH and 2,000,000 USDC, a spot price of $2,000. You swap 100,000 USDC for ETH. Using Uniswap v2’s own getAmountOut (with its 0.30% fee), the quote is 47.483 ETH. You leave the default-looking 5% slippage, so the minimum you accept is 45.109 ETH.

A searcher computes the largest front-run that still lets your trade clear at that minimum:

Now change only your slippage setting, same pool, same trade:

The attack scales with your tolerance, almost one for one. And the searcher still has to pay gas and outbid other searchers for the slot, so below a certain profit the sandwich is not worth running at all. That is why tight slippage on large trades matters more than any other setting.

What has actually happened

The bots that got robbed (April 2023). On April 3, 2023, a newly created validator exploited a flaw in the MEV-Boost relay: it signed a deliberately invalid block header, the relay revealed the block’s contents anyway, and the validator replaced it with its own block that front-ran sandwich bots’ bundles. Flashbots’ post-mortem puts the loss at about $20 million; CertiK and PeckShield’s figure was about $25 million across five bots in block 16,964,664. Relays were patched so they no longer return a block body to a proposer when publishing fails.

jaredfromsubway.eth. EigenPhi’s May 2023 analysis of the most prolific sandwich bot counted about 238,000 attacks and more than 106,000 victims from late February to May 8, 2023, with $40.65 million in revenue, $34.35 million paid in gas and $6.3 million in profit. Other analysts put its real profit lower, at $3.5 to $4.5 million; extraction figures depend heavily on methodology, so treat any single number as an estimate.

The trend on Ethereum. According to EigenPhi data published by Cointelegraph Research in December 2025, monthly sandwich extraction on Ethereum fell from nearly $10 million in late 2024 to about $2.5 million by October 2025, with average profit per sandwich just above $3 and roughly 70% of attacks linked to the same “Jared” operation. The researchers attribute part of the decline to wider use of MEV protection.

Solana. A Helius report covering December 7, 2024 to January 5, 2025 found one program executing 1.55 million sandwich transactions in 30 days for a profit of 65,880 SOL; Jito estimated nearly half of all Solana sandwiches came from it.

Not every chain works like Ethereum

Whether a sandwich is even possible depends on who can see your transaction before it is ordered.

The practical upshot: on a chain without a public mempool, a third-party bot cannot see your pending swap to sandwich it, and Ethereum-style private RPCs do nothing extra there. Your exposure on those chains comes from the sequencer or block producer, which you are trusting either way, and from slippage settings that leave room for anyone who lands next to you.

What protection actually covers

And the settings no tool replaces:

1. Keep slippage as low as the trade allows. Uniswap’s automatic setting ranges from 0.5% to 5% depending on network cost and size; for a large trade, set it yourself. Too low and the swap reverts, which costs only gas.
2. Trade in deep pools. Uniswap’s own advice for avoiding sandwiches is “Use large liquidity pools.” Smaller price impact leaves less room; market cap vs. liquidity shows how to size a trade against a pool.
3. Break up large trades on thin pools. Smaller swaps each move the price less and offer less to extract; this is common practice, not a guarantee.
4. Prefer a protected route by default on Ethereum, and check what your wallet already does on the chain you use.

Checklist before a large swap