Why bridges exist
Blockchains are built in isolated environments with different rules, so they cannot natively communicate and tokens cannot move freely between them. ETH on Ethereum and a token called ETH on another network are separate records; a bridge is the protocol that keeps them linked.[1]
Bridges carry two kinds of payload: tokens, and arbitrary messages that let a contract on one chain trigger an action on another. Chainlink CCIP, for example, supports arbitrary messaging, token transfers, and programmable token transfers that combine both.[5]
The same problem applies between Ethereum and its own rollups, even though they rely on Ethereum for security. A rollup's canonical bridge contract is what lets ETH deposited on Ethereum appear on the rollup and move back again.[3]
From a user's point of view, a bridge transfer is two transactions on two chains. Each has its own transaction hash, and the bridge's job is to connect them reliably.
Four ways to move an asset
Lock and mint: the original token is locked in a contract on the source chain and a wrapped version is minted on the destination. Burning the wrapped token later releases the original.[2]
Burn and mint: the token is burned on the source chain and minted on the destination.[2]
Atomic swaps and liquidity networks: you swap your asset on the source chain for an asset on the destination chain with another party, so the same token never crosses.[2]
The last design matters in practice for rollup exits. ethereum.org describes liquidity providers that pay a user on L1 right away and take over the user's pending L2 withdrawal, collecting it when the challenge period ends.[3]
The model decides what you actually hold at the end. With lock and mint you hold a token backed by assets locked on another chain. With burn and mint you hold a newly minted token on the new chain. With a swap-based network you hold whatever asset the other party paid you.[2]
See also: Across Protocol · deBridge · Allbridge
Trusted vs trustless
ethereum.org splits bridges by who verifies the transfer. Trusted bridges are externally verified: a federation with a multisig, a multi-party computation system, or an oracle network signs off that tokens were locked on the source chain.[2]
Trustless bridges rely on the blockchains they connect and their validators, and they do not add new trust assumptions beyond those chains.[2]
Native rollup bridges fit the trustless case. An optimistic rollup's canonical bridge releases funds on Ethereum only after the challenge period of roughly seven days; a ZK rollup's bridge can release them once the validity proof is verified on L1.[3]
Trust is a spectrum. In a hypothetical bridge with a 5-of-9 signer set, security depends on how independent those keys are; if one operator controls several, the effective threshold is lower than it looks.
Messaging protocols add their own verification design. Chainlink's CCIP documentation says cross-chain transactions use multiple decentralized oracle networks and adds configurable rate limits to cap risk. Reading that section tells you who approves a transfer.[5]
| Type | Who you trust | Typical speed |
|---|---|---|
| External validator or multisig | The signer set | Usually fast |
| Liquidity network | The counterparty or liquidity provider | Fast; provider fronts funds |
| Native optimistic rollup bridge | Ethereum plus one honest watcher | About 7 days to L1 |
| Native ZK rollup bridge | Ethereum plus proof system | After proof verification |
Worked example: moving 1 ETH off Arbitrum
You hold 1 ETH on Arbitrum and need it on Ethereum. Path one is the native bridge: you pay L2 gas to start the withdrawal, wait out the challenge period of roughly a week, then pay Ethereum gas to claim it. You trust only Ethereum and the rollup contracts.[3]
Path two is a liquidity provider. It sends you, for example, 0.998 ETH on Ethereum without the challenge wait and later collects your 1 ETH through the native route. The 0.002 ETH difference is an illustrative fee we chose for the example, not a real quote; actual fees vary with gas prices and pool depth.[3]
The second path is faster but adds trust in the provider's contracts. For a 1 ETH transfer the fee is small; for a treasury-sized transfer, the week-long native route avoids that extra layer of risk.
In both paths, the Arbitrum-side and Ethereum-side steps are separate transactions with separate hashes. If anything goes wrong, those two hashes are what support will ask for.
Where bridges fail
ethereum.org's developer docs state that bridges account for the top three biggest hacks in DeFi. Its user guide lists smart contract risk and technology risk for all bridges, plus censorship and custodial risk for trusted designs.[2]
A wrapped token is only as good as the lock behind it. ethereum.org warns of systemic risk from wrapped assets, which have been exploited: if the source-chain contract is drained or the signers are compromised, the wrapped tokens lose their backing even though nothing happened on the destination chain.[2]
ethereum.org also describes technology risk, such as software failure and buggy code, and censorship risk: operators of a trusted bridge can theoretically stop users from transferring assets. It cites the Wormhole hack, in which 120k wETH (about $325 million) was stolen.[1]
Token approvals add a separate risk. Bridges often ask for an approval to spend your tokens; an unlimited approval to a bridge contract that is later exploited can drain more than you bridged. Approving only the amount needed limits that exposure.
See also: Revoke.cash · Token approval in the glossary
Checklist before you bridge
Confirm the route. For Ethereum and its rollups, use the rollup's native bridge, such as the Arbitrum Bridge that ethereum.org cites; for other chains, check which bridge the token issuer itself uses.[1]
Confirm the destination asset. Look up the token contract on the destination chain and check that it is the one wallets and DEXs there actually trade, not a lookalike wrapped version.
Send a small test first. A transfer of a few dollars shows the real fees, the wait time, and whether the asset arrives where you expect, before you commit the full amount.
Save both transaction hashes. The source-chain hash proves you sent the funds; the destination-chain hash proves they arrived. Support teams trace stuck transfers with these two IDs.
Revoke leftover approvals afterward. Once the transfer completes, the bridge no longer needs permission to spend your tokens, and removing it closes one path an exploit could use.
See also: Self-custody in the glossary
The bottom line
Choose a bridge by what backs the asset you receive. For moves between Ethereum and its rollups, the native bridge carries no extra trust and costs you time; fast bridges save the week for a small fee and add contract risk. For any bridge using an external signer set, the signer set is the security, so size your transfer to what you would accept losing if it failed.[2]
Educational content, not financial advice. Crypto assets are volatile; do your own research.