Every swap has two moments: the quote you see and the settlement on-chain. Between them, other transactions land, pool reserves shift and the price you were shown can drift. Slippage is that drift, measured at settlement. It is normal, it is usually small and it becomes manageable once you know what moves it.
Why do quoted and settled rates differ?
Your transaction does not settle the moment you click. It waits for a block, and blocks arrive about every 12 seconds on Ethereum, so any swap that lands ahead of yours in the same pool changes the reserves your rate is computed from. In a deep, calm market the drift usually rounds to zero. In a thin or fast one, it can decide whether the fill is one you wanted.
Two forces set the size of the drift. Volatility raises the odds that the market moves while you wait, and low liquidity raises how much every trade ahead of yours moves the price. The same 30-second wait can be invisible on a major pair and expensive on a small one.
Where does the painful kind live? Mostly in small pools and fast moments: a token minutes after launch, a pair repricing on news, a pool drained thin by an earlier trade. On major pairs, meaningful slippage is rare outside genuine volatility spikes.
What is price impact, and how is it different?
Price impact is the part of the difference you cause yourself. An automated market maker prices a swap from pool reserves, so your own order moves the price as it fills; that cost is knowable in advance and is already inside your quote. Slippage is the part the market causes while you wait, and it is only knowable at settlement. Your tolerance setting covers the second part.
The scale of price impact is mechanical. In a pool holding 1,000 ETH and 3,000,000 USDC, swapping 10 ETH, 1% of the pool's ETH, delivers about 29,700 USDC: an average rate about 1% below spot. Push 100 ETH through the same pool and the average lands 9.1% below spot. Aggregation attacks exactly this cost: splitting the order across venues keeps each slice small against its pool, which is part of how 1inch quotes competitive rates, aggregated across 300+ sources.
The two costs also respond to different levers. You reduce price impact by trading smaller relative to the liquidity, by splitting across venues or by picking deeper pools. You reduce slippage exposure by quoting fresh, settling fast and keeping the window between quote and confirmation short. A good interface works both levers for you; the tolerance is the one dial it leaves in your hands.
How does slippage tolerance work?
Tolerance converts into a minimum received amount written into the transaction itself. If the swap cannot deliver at least that amount, it reverts: you pay the network fee for the attempt, but no tokens change hands. Set 0.5% on a swap quoted at 3,000 USDC and it can settle anywhere down to 2,985 USDC, and nowhere below.
Minimum received is the number to actually read before signing. It folds the quote, your tolerance and the route's price impact into the one figure the contract will enforce, so it answers the only question that matters: what is the worst case I am agreeing to? If that number is acceptable, the setting is right.
- 0.1% or lower
- For stable pairs and deep pools. Expect reverts if you point it at a volatile pair.
- 0.5%
- A common choice for liquid pairs in normal conditions.
- 1% or higher
- Fills through volatility, and widens what a bad fill can cost. Use it deliberately, not by habit.
- Auto
- 1inch picks a tolerance for the pair and current conditions, tightening it whenever the market allows.
What happens when tolerance is set too high?
A high tolerance does more than accept accidental drift: it announces, on the public record, how bad a fill you are willing to sign. Sandwich bots are built to read exactly that. They push the pool's price toward your limit, let your swap fill there and unwind right after, keeping the difference.
The numbers say this is routine, not exotic. EigenPhi counted more than 95,000 sandwich attacks on Ethereum between November 2024 and October 2025, and almost 40% of them hit stablecoin pools, the pairs traders assume cannot move. Extraction is falling as protections spread, from about $10 million a month in late 2024 to about $2.5 million by October 2025, but the attack itself has not gone anywhere. Intent-based swaps on 1inch avoid the exposure at the root, because the order never waits in the public mempool.
Good defaults beat vigilance here. A tolerance you set once and forget will eventually meet a market it does not fit; a mode that recomputes it per swap will not. That is the case for auto-slippage on routine swaps, and for treating any manual increase as a decision you can explain.
How do you choose the right tolerance?
- 1
Start from the pair
Stable pairs and majors in deep pools sit comfortably at 0.1% to 0.5%. Long-tail tokens in shallow pools genuinely need more room.
- 2
Check your size against the pool
If your order is a visible fraction of the pool's liquidity, expect price impact and consider splitting the swap or sizing it down.
- 3
Mind the moment
News, listings and liquidations widen drift in minutes. A setting that was right yesterday can be wrong during a move.
- 4
Let auto mode carry the routine
Auto-slippage on 1inch adjusts the tolerance to the pair and conditions, so manual settings are for exceptions, not defaults.
What should you read next?
A tolerance set too high is exactly what sandwich bots hunt for. Read how MEV protection works and what a DEX aggregator does, or browse all Learn guides.
Frequently asked questions
Not in a moving market, and no honest quote promises it. What can be fixed is the floor: the minimum received in your transaction is binding, and an intent-based order on 1inch settles at the rate you signed or better, or not at all.
The market moved past your tolerance before the transaction confirmed. A revert costs the network fee but swaps nothing. Requote and try again, and treat repeated reverts as information about size or timing rather than a reason to crank the setting up.
Usually not: drift is how open markets behave. The pattern that deserves suspicion is a fill landing exactly on your tolerance limit on a quiet pair, which is the signature of a sandwich attack. MEV protection exists for precisely that case.
No. A limit order fills at the price you set or it does not fill, which makes it the better tool when the exact rate matters more than timing. On 1inch, limit orders also need no native token for gas.
Swap with the drift under control
Auto-slippage, a binding minimum received and competitive rates, aggregated across 300+ sources.
