Layer 2 Withdrawal Times: Up to Ten Days Between the Finalized Stamp and Your Money

When you move funds from a Layer 2 network back to Ethereum, your wallet tells you after roughly twenty minutes that the process has been finalized. Your money has still not arrived on Ethereum at that point, and depending on the network it takes anywhere from just under two hours to more than ten days afterwards. This gap between what the display says and what is actually happening is why withdrawals started shortly before an exchange deadline regularly arrive too late.

I measured the waiting times myself on September 14, 2026, directly against the contracts on Ethereum and against the public nodes of six Layer 2 networks. The short answer: on the optimistic networks Arbitrum One, OP Mainnet and Base, the finalized marker sits some eighteen to twenty-four minutes behind the current state, yet the waiting period written into the contract runs to between 1 and 7 days. On the ZK networks zkSync Era, Linea and Scroll, the marker lags by 1.7 to 4.1 hours, and there it roughly describes the moment your funds become claimable on Ethereum. Same term, two entirely different meanings.

Layer 2 withdrawal explained: what really happens on the way back to Ethereum

A Layer 2 is a network of its own. It processes its transactions itself and then posts the results to Ethereum in batches. The benefit shows up in the cost; the price is paid at withdrawal. Anyone pulling funds back has to wait until Ethereum has accepted the Layer 2 result as valid.

The route has three steps, and only the first one is quick. You begin by starting the withdrawal on the Layer 2 itself. The network then has to submit the corresponding state to Ethereum, and only after that does the real waiting period begin. At the end comes a second transaction on Ethereum, the one with which you claim the funds. Anyone who completes only the first step and then waits may be waiting indefinitely, because on most networks the claim does not happen by itself.

State root, proof and claim

A state root is a single check value that summarises the entire balance sheet of a Layer 2 network at one point in time. That check value is the anchor against which Ethereum later verifies whether your withdrawal really belongs to a valid state. As long as no state root has been posted to Ethereum for the block containing your withdrawal, you cannot prove anything at all, however long you wait.

Only once the root is in place do you submit the proof. After that, the window runs in which other participants may object. If it expires without objection, you claim the money.

The finalized stamp: why "finalized" in the block explorer means two different things

Every node on an EVM network knows a block marker called finalized. That marker describes the state the network regards as irreversible. On Ethereum itself this is unambiguous. On a Layer 2 the meaning depends on how the network is built, and this is precisely where the misunderstanding arises.

On the optimistic networks, the marker refers to the data from which the Layer 2 state is derived. Once that data is final on Ethereum, the Layer 2 block derived from it counts as finalized. That says nothing about your withdrawal. The fraud window keeps running independently.

On the ZK networks, the marker is tied more closely to what actually matters to you: there a batch is only carried as final once the corresponding proof has been verified and executed on Ethereum. Anyone reading their withdrawal out of that marker on Arbitrum or Base is off by a factor of several hundred.

Our own measurement: what the contracts of six Layer 2 networks prescribe as withdrawal periods

This analysis was carried out by cryptoticker.io on September 14, 2026. The measurement was taken at 06:56 UTC against Ethereum block 25,974,039, via a public node, with no account and no keys. In each case I queried the parameters that sit in the contract on Ethereum and set the waiting period.

One methodological detail makes the difference: I did not take the contract addresses from a list but resolved them starting at the canonical bridge contract. On Arbitrum One, the bridge leads to a different rollup contract from the one many older guides name; the older one reports a confirmation state from February 2025 and has therefore been superseded. Query the wrong address and you get an answer that looks like a measurement.

The maturity period is the time that has to elapse between your submitted proof and the permitted claim. The lock period is an additional wait that the operator has written into the contract as a safety buffer. Both values sit in the same contract. Whether they add up or overlap in an emergency depends on how the dispute procedure runs, and I did not verify that with a withdrawal of my own. Anyone planning conservatively adds them together.

Optimistic rollups: why the fraud window on Arbitrum runs to 6.4 days

An optimistic rollup assumes the submitted results are correct and grants everyone else a window in which they may prove the opposite. That window is the fraud window, and it is the real reason for the wait. It is sized so that an honest participant can still object even if someone tries to push them off the network for a while.

On Arbitrum One, the rollup contract holds a value of 45,818 Ethereum blocks. Converted at the target time of 12 seconds per block, that gives 549,816 seconds, or 6.36 days. The conversion is the only place in my measurement where an assumption is buried: Ethereum blocks arrive on a twelve-second rhythm on average, and individual slots can be missed. In practice that lengthens the window rather than shortening it.

On top of that comes a grace period of 14,400 blocks, around two days. In normal operation this grace period does not apply; it becomes relevant only if there is an actual dispute. For your planning that means 6.4 days is the standard case and 8.4 days is the upper bound you should work with if you have no buffer.

OP Mainnet and Base compared at contract level: 7 days against 1 day on the same technology

The most interesting finding of the measurement sits between two networks running on the same software. OP Mainnet carries a maturity period of 604,800 seconds in its portal contract, exactly seven days, plus a lock period of 302,400 seconds, or three and a half days. Base carries 86,400 seconds in the identically built contract, so one day, and a lock period of zero.

Both values come from the same query at the same moment, and both networks run on the same software. The difference is therefore a decision taken by the respective governance, not a technical necessity. That is also why you should not rely permanently on a figure you read once: what stands at one day today can be back at seven after a contract update. The number sits publicly in the contract and can be looked up at any time.

What this means for the network you choose

If you regularly move back and forth between a Layer 2 and Ethereum, withdrawal duration is a hard selection criterion alongside fees. The duration determines how quickly you can react to a cut-off date, such as a withdrawal deadline at your crypto exchange. Anyone simply holding funds for the long term never feels the difference. Anyone working with them feels it every time.

ZK rollups: why zkSync, Linea and Scroll settle in hours rather than days

A ZK rollup does not present Ethereum with a claim that would have to be contested, but with a mathematical proof that the contract itself recomputes. If the proof passes, the state is valid. No fraud window is needed, because there is nothing to challenge.

The remaining wait arises because a proof is always generated for whole batches of blocks and generating it costs computing time. That lag is exactly what I measured, by querying each network for its current block and its block carried as final, then comparing the timestamps.

At first glance the table reads the wrong way round, and that is its value. The three optimistic networks sit at the top because their marker shows the least lag, even though it is precisely there that the longest real wait is coming for you. The three ZK networks sit at the bottom, even though their figure is the only one that gives any indication of when you get to your money.

Submission rhythm measured: every 60 minutes on OP Mainnet, every 19 minutes on Base

Ahead of the fraud window sits a wait that most guides leave out: your withdrawal can only be proven once the state containing it has been submitted to Ethereum at all. To gauge this, I read out the 24 most recent submissions for OP Mainnet and Base via the relevant contract and compared their timestamps.

On OP Mainnet the median is 60.4 minutes, with a range of 60.2 to 60.4 minutes. The window ran from September 13, 07:23 UTC to September 14, 06:30 UTC. The rhythm is therefore effectively hourly and barely fluctuates. On Base the median is 19.2 minutes, but the range runs from 4.4 to 35.2 minutes, measured from September 13, 23:08 UTC to September 14, 06:43 UTC. Base submits more often, but less regularly.

For planning purposes you add this time on top. On OP Mainnet, waiting for the next submission alone can cost you a full hour, on Base up to a good half hour. Set against a seven-day window, that barely registers. If you start a withdrawal on the last possible day, that hour is what decides it.

Layer 2 withdrawal before an exchange deadline: how to count backwards

The practical occasion is on the table right now. KuCoin has delisted 25 tokens and closes withdrawals on October 7, 2026 at 08:00 UTC; we have listed the withdrawal deadline and the affected tokens individually. Anyone who still has to pull funds out of a Layer 2 for a cut-off date like this and then send them to an exchange is best advised to count from the end.

For October 7, 08:00 UTC, the measured values give the following latest start times, in each case without a buffer and without the crediting time at the receiving exchange:

• OP Mainnet: 7.00 days of maturity period plus 3.50 days of lock period give 10.5 days on a conservative reading. Latest start: September 26, around 20:00 UTC.
• Arbitrum One: 6.36 days of fraud window plus 2.00 days of grace period give 8.36 days. Latest start: September 28, around 23:20 UTC.
• Base: 1.00 day of maturity period. Latest start: October 6, 08:00 UTC.
• Scroll, Linea, zkSync Era: 1.7 to 4.1 hours. The day before is enough.

A buffer belongs on top of these values, for four reasons: the wait until the next submission, the crediting time at the receiving exchange, possible network congestion, and the plain fact that you have to trigger the second transaction yourself. Anyone who starts over a weekend and only notices the claim on Monday loses two days that appear in no contract period.

The two transactions many people forget

On the optimistic networks it is usually two signatures on Ethereum: one for the proof, one for the claim. Both cost fees on Ethereum, not on the Layer 2. So keep enough ether ready on the address you are withdrawing from. What ether is currently worth is shown on our Ethereum page. A withdrawal left hanging on an empty gas balance waits beyond the seven days, and goes on waiting until you top it up.

Fast bridges as a shortcut: what you take on in exchange for the time saved

There is a route that bypasses the wait. So-called fast bridges pay you the funds out on Ethereum immediately and sit out the waiting period themselves. They charge a fee for this, and you trade a wait for counterparty risk.

That risk is not theoretical. cryptoticker.io reported on the outflow of funds at the Symbiosis bridge on September 12, 2026, and on the shutdown of the Silicon Network bridge, with a deadline of its own, on September 4, 2026. The canonical bridge of a Layer 2, by contrast, is the one operated by the network itself; it is slow, but it has no counterparty that can disappear.

That makes the trade-off an honest one to weigh. For small amounts and a tight deadline, the fast bridge can be the right choice. For amounts whose loss would hurt, the wait is the price of safety, and the moment to pay it is before the cut-off date.

What the measurement does not show: the limits of this analysis

The values come from the contracts and from the nodes, not from a withdrawal of my own with a stopwatch. What I can evidence are the periods the network prescribes and the rhythm in which submissions are made. What I cannot evidence is the actual duration of a specific withdrawal from start to finish.

Further open points you should be aware of: the submission rhythm rests on a single day's sample of 24 submissions per network, not on a long-term average. The marker for the final state is set by each node provider individually, and I queried exactly one public endpoint per network. Layer 3 networks, non-EVM chains and all third-party bridges are not measured. And whether the maturity period and the lock period on OP Mainnet add up in an emergency is the conservative reading, not an established fact.

The technical foundations of both designs are publicly documented, at Ethereum itself for optimistic rollups and in the developer documentation of the OP Stack networks for the course of a withdrawal. Anyone wanting to recompute the figures in this article will find there the contract names I queried.

Planning a Layer 2 withdrawal: what to take away

1. Count backwards, not forwards. Put the cut-off date at the start and subtract your network's period: 10.5 days on OP Mainnet, 8.4 on Arbitrum One, 1 day on Base, a few hours on the ZK networks. If the cut-off comes from an exchange, check at the same time whether your exchange supports the network directly at all; which providers are licensed in the EU is set out in our overview of regulated crypto exchanges.
2. Do not rely on the finalized stamp. On the optimistic networks it describes something other than your withdrawal. What governs is the contract value, and that can be queried publicly. Anyone holding funds in self-custody should plan firmly for the second transaction; the equipment for that is covered in our hardware wallet comparison.
3. Document the process while it is running. A withdrawal spread over several days produces two transactions on Ethereum and one on the Layer 2, and come January you will want to know which of them belong together. A portfolio tracker takes the piecing-together off your hands.

(As of September 14, 2026. This article is not investment advice. Prices and fee structures change; check the terms with the provider before you buy.)