MEV on Uniswap: How Sandwich Attacks Work and Strategies to Minimize Front-Running

A trader submits a swap on Uniswap expecting to exchange 10 ETH for USDC at a particular price. Within seconds, the transaction sits in the mempool—visible to node operators and validators before it is executed on-chain. Another actor observes this pending transaction, inserts their own swap ahead of it to move the price unfavorably, and then inserts a third transaction behind the original swap to capture the slippage. The trader receives less USDC than anticipated, while the attacker profits from the difference. This is a sandwich attack, a concrete example of maximal extractable value, or MEV, and it represents one of the most persistent costs embedded in DeFi protocols.

Uniswap’s decentralized exchange architecture removes custodians and order books, but it does not eliminate the information asymmetry that allows sophisticated actors to extract value from ordinary traders. The automated market maker model that powers Uniswap—where liquidity pools and algorithmic pricing replace traditional market makers—actually creates new opportunities for extraction because all pending transactions are temporarily visible and all price movements are deterministic and observable. Understanding how these attacks work and what defensive mechanisms exist is now a required skill for any serious DeFi participant.

How the mempool and transaction ordering create opportunity for extraction

Every transaction submitted to the Ethereum network enters the mempool, a waiting area where it remains visible to validators, node operators, and network participants before being included in a block. For a Uniswap swap, this means the token pair, input amount, expected output, and wallet address are observable to anyone monitoring the network. Validators and block producers have extraordinary power in this environment: they decide which transactions enter a block, in what order, and therefore which prices traders will face when their swaps execute.

The automated market maker mechanism makes this visibility especially dangerous. Uniswap prices tokens algorithmically based on the ratio of assets in a liquidity pool. If a large swap is pending, the protocol will execute it against the pool at whatever price results from the constant product formula (or other mechanisms in V3), meaning the final price is entirely predictable once the transaction size and pool state are known. An attacker observing the pending swap can calculate exactly what price the trader will receive and whether there is profitable space to extract value through reordering.

On-chain finality also matters. Ethereum transactions become final once included in a block and confirmed by subsequent blocks. But the path from mempool to finality passes through a critical moment: the builder’s choice about ordering. Proposer-builder separation (PBS) and MEV-Boost have made this dynamic more visible. Builders now compete to construct blocks that maximize extractable value, and sandwich attacks are among the most reliable sources of that value. A builder can observe a pending Uniswap swap, include a front-running transaction, the victim’s swap, and a back-running transaction all within a single block, guaranteeing execution and profit.

Sandwich attacks: The three-transaction exploit and profit mechanics

A sandwich attack proceeds in three steps. First, the attacker inserts a transaction that moves the price in a direction harmful to the victim’s swap. For a swap of ETH to USDC, this means buying ETH or selling USDC in the liquidity pool, raising the ETH price and lowering the USDC received. Second, the victim’s transaction executes at this worse price, unaware that the pool state has changed. Third, the attacker executes a back-running transaction that reverses the initial price movement, typically by selling the asset they front-ran with, pocketing the difference between their entry and exit price.

The mathematics of the exploit are straightforward. Suppose a trader submits a swap of 100 ETH for USDC when the pool price is 3000 USDC per ETH. An attacker observes this transaction and front-runs with a purchase of 20 ETH, raising the price to approximately 3300 USDC per ETH (the exact price depends on pool depth and the constant product formula). The victim’s 100 ETH now buys roughly 330,000 USDC instead of 300,000—a loss of about 70,000 USDC. The attacker then back-runs by selling their 20 ETH at the new price, capturing most of the slippage the victim experienced. After accounting for gas costs, the attacker profits while the trader loses.

The attack scales with transaction size and pool liquidity. Large trades moving the price further create larger profit opportunities. Shallow liquidity amplifies the effect. The attacker’s profit is also limited: if the pool is very deep or the victim’s trade is small, the MEV opportunity shrinks below the cost of the attack (typically two to three transactions at current gas prices). But for significant trades against moderately liquid pools, sandwich attacks routinely extract hundreds or thousands of dollars.

Why Uniswap V3 concentrated liquidity increased MEV exposure

Uniswap V3 introduced concentrated liquidity, allowing liquidity providers to focus capital within specific price ranges rather than distributing it across the entire price curve. This innovation improved capital efficiency for providers and reduced slippage for small trades—but it also increased price impact for large swaps. With less total liquidity deployed across the full range, a substantial trade can move the price further and faster, creating larger MEV opportunities.

Concentrated liquidity also created new extraction patterns. MEV searchers can now identify liquidity «gaps»—price ranges where liquidity is thin—and target trades that would cross multiple concentrated positions. The Uniswap V3 architecture, with its tick-based price grid and custom liquidity curves, is mathematically transparent. Given a swap size and current pool composition, an attacker or searcher can calculate exactly how far the price will move and what profit is available.

The trade-off reflects a broader MEV truth: mechanisms that improve capital efficiency or reduce certain frictions often introduce new extraction opportunities elsewhere. V3’s efficiency gains are real, but they came at the cost of a new attack surface. Traders using Uniswap V3 must now account for the fact that their large swaps will move prices further and create more visible MEV opportunities than they would have in V2.

Flash swaps and other advanced MEV mechanics

Uniswap’s flash swap feature allows a user to receive tokens from a liquidity pool without holding the full payment upfront, provided they return the equivalent amount (plus fees) within the same transaction. This was designed to enable efficient arbitrage and liquidation flows, but it also became a tool for MEV extraction. An attacker can flash-swap a large amount, execute a series of other swaps to profit from price movements or inefficiencies, and then repay the flash loan from the proceeds.

These flash-swap-enabled attacks can be complex. An attacker might flash-swap tokens to manipulate the price of a liquidity pool on Uniswap, force other traders’ transactions to execute at worse prices, and then repay the flash loan from extracted MEV. Or they might use flash swaps to arbitrage price differences between Uniswap and other decentralized exchanges, capturing value that would otherwise be available to casual traders.

Time-weighted price oracles, which Uniswap provides natively, also play a role in MEV dynamics. Because the oracle is built into the smart contracts, other protocols rely on it for pricing. An MEV searcher who can temporarily manipulate Uniswap’s pool composition can cause the oracle to report incorrect prices, triggering liquidations or favorable trades on protocols that depend on it. This is a second-order extraction: the direct MEV on Uniswap itself is secondary to the extraction enabled on dependent protocols.

Private mempools and encrypted transactions: Partial defenses

The most direct defense against sandwich attacks is to hide the transaction from public observation until it executes. Private mempools accomplish this by allowing traders to submit swaps directly to a private pool of validators or block builders rather than broadcasting them to the public network. Services using Flashbots MEV-Protect and similar encrypted transaction protocols send transactions encrypted to relayers who decrypt and include them in blocks without revealing them to searchers in the public mempool.

These tools reduce sandwich attack risk significantly. A private mempool transaction is invisible to other MEV searchers, eliminating the information asymmetry that makes sandwich attacks profitable. If thousands of traders use private pools, the coordinated attack surface shrinks because large public transactions become rarer.

However, private mempool solutions introduce their own trade-offs. They require trust in the relay infrastructure: a misbehaving relay operator could steal funds, front-run transactions themselves, or sell order flow to the highest bidder (a legal but ethically contested practice called MEV sales). Some private pool operators have faced criticism for this exact behavior. Additionally, privacy pools may introduce latency or reduce transparency about actual execution—you send in an encrypted transaction and receive a promised output, but the actual mechanics remain opaque until settlement.

Slippage tolerance, limit orders, and on-chain execution guarantees

For traders without access to private relays, slippage tolerance is the most accessible tool. Uniswap’s router contracts allow a user to set a minimum acceptable output: if the actual swap would produce less than this amount, the transaction reverts, and no value is extracted. A trader protecting themselves might set slippage tolerance at 0.5% instead of 1%, accepting a higher rejection rate to catch sandwich attacks that push the price too far.

The limitation is that slippage tolerance is reactive, not preventive. The trader still submits a public transaction, still gets front-run and back-run, but merely refuses to complete the unfavorable execution. A sandwich attack simply fails and can be retried. More sophisticated traders use private relays or mev-resistant approaches rather than relying solely on slippage limits. However, slippage tolerance remains the default protection for ordinary users and remains effective at preventing the worst-case extraction.

Limit orders are another improvement, allowing a trader to specify a price rather than a size and accepting whatever amount can be purchased or sold at that price or better. Limit orders reduce information leakage because the exact quantity being traded becomes uncertain until execution. An attacker front-running a limit order does not know whether the resulting price movement will cause the order to fill at all. Several protocols have implemented intent-based limit orders on top of Uniswap or other DeFi protocols, though these introduce new dependencies and costs.

MEV-resistant smart contracts and DeFi protocols, discussed across sites.google.com/cryptowalletextensionus.com/uniswap/ and other technical resources, aim to guarantee that a swap will execute at a particular price regardless of ordering, or will fail if that price cannot be met. These are more sophisticated than simple slippage tolerance and may offer better execution guarantees for large trades, though they require explicit opt-in and may have higher fees.

Batch auctions and protocol-level MEV solutions

Some of the most promising defenses operate at the protocol level rather than the application layer. Batch auctions group pending transactions into bundles and execute them together, eliminating ordering-based extraction because the order within a batch is randomized or does not matter. CoW Swap, built on top of Uniswap and other liquidity sources, uses batch auctions: traders submit their intent to swap, the protocol collects these intents into batches, solvers submit execution bundles, and transactions execute together. An attacker cannot insert a transaction ahead of a victim’s swap because all transactions in the batch settle simultaneously at the same clearing price.

The trade-off is liquidity fragmentation and latency. Batch auctions settle periodically, not continuously, and may accept slightly worse prices than real-time execution if competition among solvers is insufficient. But for traders concerned with MEV, the guarantee of no sandwich attacks can justify the cost. Batch auctions also encourage better price discovery because solvers compete to find the best execution across multiple liquidity sources.

Encrypted mempools at the protocol level—where validators themselves cannot see pending transactions until they are included in a block—represent a more radical defense. This approach requires consensus-layer support, as evidenced by research on encrypted transactions in Ethereum’s roadmap and alternative chains experimenting with threshold encryption. These solutions could eliminate sandwich attacks entirely at the source but introduce complexity and require changes to core protocol infrastructure.

The realistic cost of MEV protection and choosing tools

Every MEV defense carries a cost. Private relays may charge fees or sell order flow. Batch auctions introduce latency or worse prices. Slippage tolerance may cause transaction failures. Limit orders may not fill immediately. Understanding these trade-offs allows a trader to choose protection proportionate to the value at stake. A $500 swap probably does not justify a 0.3% private relay fee, but a $50,000 trade might easily save more in MEV costs than the relay fee costs.

For retail traders and small-to-medium swaps, slippage tolerance set appropriately—typically 0.3% to 1% depending on market volatility and liquidity—remains the most practical tool. Combined with awareness of transaction size and pool liquidity, this catches the most egregious sandwich attacks while keeping execution simple. For traders executing large positions, private relays, batch auctions, or limit orders are worth evaluating. MEV extraction is not purely an academic concern; sophisticated searchers extract billions annually. But ordinary traders also have effective defenses available if they understand the mechanics and choose the tools suited to their risk profile.