Anyone moving a balance off a crypto exchange and into their own wallet pays a fixed amount for it, which the exchange calls a withdrawal fee. How much that is sits in the small print of each platform. What the same transfer actually costs on the network does not. We measured both numbers on August 17, 2026 and set them side by side.
The result is starker than expected. For a Bitcoin withdrawal, Bitvavo charged 0.000023 BTC on that day, or €1.25. An ordinary Bitcoin transaction cost between €0.08 and €0.15 in the same minute. That puts the factor somewhere between eight and sixteen. Across all 429 assets we could price, the withdrawal fee sat at €0.84 in the middle of the distribution, and for 158 of them it fell between 90 cents and one euro.
This analysis was carried out by cryptoticker.io on August 17, 2026.
A crypto withdrawal consists of two separate operations that show up on your statement as a single number. First the exchange has to write a transaction to the network in question. That carries a fee which goes to miners or validators, not to the exchange. On Bitcoin it is priced by the size of the transaction in virtual bytes and by how busy the network is, not by the amount being sent.
On top of that sits the exchange's own withdrawal fee. This amount is a price the company sets. It can pass the network cost through exactly, it can exceed it, and in rare cases it can undercut it where a platform subsidises withdrawals. As an investor, only the second number is deducted from your balance, so only that one matters directly. The first is the yardstick that tells you whether the price is reasonable.
The difference turns practical the moment a deadline is bearing down on you. Several providers currently have cut-off dates by which balances have to be moved out. Anyone who starts doing the maths at that point is doing it under time pressure. The tax side of such a transfer is a separate question: whether the network fee counts for tax purposes when you send between your own wallets does not depend on what the exchange charges on top.
The basis is three publicly available data sources, which we called in sequence within a few seconds at around 00:56 UTC on August 17, 2026. First, the asset list from Bitvavo, an exchange licensed in the Netherlands by the AFM under MiCA that serves German customers directly through the EU passport. That list carries, for every asset, the withdrawal fee, the minimum withdrawal amount, the status and the network. Second, the price list from the same exchange, which we used to convert the fees into euros. Third, the fee recommendation and mempool state from mempool.space.
We checked 475 assets. For 429 of them a euro market existed, so we could convert the fee, quoted in coin units, into a euro amount. The remaining 46 were taken out of the money calculation and kept only in the status count. The conversion rate used was exclusively the price from the same exchange in the same minute, so that fee and rate do not come from two different moments.
There were four things we could not check. We did not trigger a withdrawal and therefore did not verify whether the amount billed matches the amount displayed. We have no access to the terms offered to other user groups, such as discounts for high-volume traders. We measured a single point in time, and the values move. And we could not capture the fee schedules of the other providers that matter to German investors in the same depth, because they do not offer an open API carrying fee data.
For Bitcoin the list showed a withdrawal fee of 0.000023 BTC. That is 2,300 satoshi. At the simultaneously retrieved rate of €54,271 per Bitcoin, that comes to €1.25. The minimum withdrawal amount stood at exactly the same value, so also at 0.000023 BTC.
That equality is not a coincidence but a design principle running through the entire list. The minimum is set so that it just about covers the fee. Withdraw the minimum in Bitcoin and, on paper, you end up with nothing. This is not a trap but a lower bound meant to stop tiny amounts from loading the system. For you it still means one thing: the minimum is not a sensible withdrawal size, it is the point at which a withdrawal stops being worth doing.
In the same minute, mempool.space recommended 2 satoshi per virtual byte for a fast confirmation and 1 satoshi per virtual byte for confirmation within the next hour. The mempool held 76,262 unconfirmed transactions with a total volume of around 43.6 million virtual bytes. The median fee of the block expected next stood at 1.67 satoshi per virtual byte, and at 0.54 and 0.38 in the projected blocks after it.
An ordinary Bitcoin transfer with one input and two outputs, in the SegWit format standard today, occupies around 141 virtual bytes. At 1 satoshi per virtual byte it costs 141 satoshi, or €0.08. At 2 satoshi per virtual byte it is 282 satoshi and therefore €0.15. The withdrawal fee of 2,300 satoshi corresponds, at the higher recommended rate, to a transaction of 1,150 virtual bytes, and at the lower rate to one of 2,300 virtual bytes. That is a multiple of what a single transfer requires.
Across all 429 positions we could price in euros, the picture is as follows. The bottom quarter of the values costs up to €0.30, the median sits at €0.84, and the top quarter starts at €0.95. The arithmetic mean, at €1.43, lies well above the median, which comes down to a single extreme value we return to further below.
For 271 of the 429 assets the fee is above 50 cents. Thirty-two come in above one euro, and exactly one above two euros. At the bottom end there are 18 assets with a fee below ten cents and three for which no fee is charged at all. The spread therefore runs from zero to a three-digit amount, and it follows what the respective networks cost only in part.
The most striking structure in the data is a cluster. Of the 429 fees, 158 fall between 90 cents and one euro, and a further 28 between 80 and 90 cents. Together that is 186 assets, more than four in ten, sitting inside a corridor twenty cents wide. That corridor has little to do with the networks, because the assets involved are spread across very different chains.
The finding sharpens when you look at the Ethereum network. Of the 475 assets, 241 are carried there, and their withdrawal fee has a median of €0.95 with a range from €0.40 to €1.56. Transferring an ordinary ERC-20 token imposes a similar load on the network regardless of the token, so a uniform price is understandable. What is notable is the uniformity in euros: the fee appears to be conceived as a euro amount and then converted into the relevant coin unit.
Bitvavo states in its own help centre that withdrawal costs are an estimate of network fees and can change without notice. The numbers support that account insofar as there are indeed no tariffs fixed for months at a time. They also show, however, that at the moment of measurement the estimate sat considerably above what the Bitcoin network was asking. How that estimate is arrived at in detail is not apparent from the public data, and we did not ask.
Set the fee against the minimum withdrawal amount and, across 428 usable positions, you get a median of 92.5 percent. For half of all assets, in other words, the fee eats more than nine tenths of the smallest withdrawal possible. For 281 assets it is at least half, for 228 at least nine tenths. At 26 assets the fee reaches or exceeds the minimum.
In absolute terms the minimums are small. The median sits at one euro and the highest value at €8.56, for XRP at ten units. Nobody is being kept from withdrawing by a high minimum. The real message in the numbers is a different one: withdraw your balance in many small steps and you pay the fee every single time, and on small amounts it is crushing in relative terms. On a €10 Bitcoin withdrawal, €1.25 goes on the fee, more than an eighth of the amount.
All 475 assets checked are withdrawn over exactly one network, none over two or more. In total 109 different networks appear, 96 of which carry a single asset. The large pools are Ethereum with 241 assets, Solana with 46, Base with 30 and BNB Smart Chain with 26.
For you that means an absent choice. With USDC, for instance, the route runs exclusively over Ethereum, at a fee of 1.1 USDC, or €0.95. Withdrawing the same stablecoin over a cheaper chain is not on offer here. At providers that do offer several chains the picture differs: at KuCoin, six chains were available for Bitcoin at the same moment, four of them enabled.
That comparison does not, however, favour the provider with the wider selection. For a withdrawal on the real Bitcoin network, KuCoin charged 0.00009 BTC, or €4.88, almost four times as much. The range of chains and the level of the fee are two separate qualities, and a provider can be strong on one and weak on the other. Anyone withdrawing regularly should therefore check both before settling on a regulated exchange, rather than looking only at the trading fee.
The lowest Bitcoin withdrawal fee in our cross-check stood at 0.000004 BTC, or 22 cents. It applies to the route over BNB Smart Chain. Take that route, though, and you do not receive Bitcoin on the Bitcoin blockchain but a token pegged to Bitcoin on a different chain. That token depends on a custodian and on a bridge mechanism, and both are additional risks that do not exist on the Bitcoin network.
The Lightning Network is a different matter: real Bitcoin arrives there, simply over a payment channel rather than a base-layer transaction. The fee stood at 0.00002 BTC, or €1.09. If you are moving smaller amounts anyway, this is a sensible alternative, but it requires a wallet that supports Lightning. Far from every wallet does.
Part of the inventory cannot be moved at all at the time of the survey. Of the 475 assets, 434 carried the status OK, 22 the status maintenance and 19 the status delisted. Deposits and withdrawals were always affected together; there was no asset where only one direction was blocked.
For two assets an explanatory note was attached, stating that the asset was in trading-only mode and that deposits and withdrawals were not possible. For the other 39 no reason was given. An asset sitting in maintenance can have many causes, from a network upgrade through a technical fault to a migration in progress. The API does not say which.
Cross-checks at two further venues show that a base layer of suspended assets is nothing unusual. At Bitstamp, four of 168 listed currencies were blocked for withdrawals. At Coinbase Exchange, 406 of 491 crypto entries carried the status online and the remaining 85 carried a different one. The proportions differ; the underlying pattern does not.
For you this is the practically most important point of the analysis. If you want your balance out by a certain date, it is not the fee that decides first but the question of whether withdrawals are enabled at that moment at all. That information sits in the app before you trigger the transaction, and it can change within hours.
A single entry behaves unlike all the others. For Scroll, a withdrawal fee of 19,000 SCR was on file, which at the retrieved rate comes to €329.48. The minimum withdrawal amount for the same asset stood at 290 SCR, or €5.03. The fee therefore exceeds the minimum by a wide multiple, which does not resolve arithmetically: a withdrawal at the minimum could not cover the fee.
We carried this value through unchanged, because that is how it stood in the public API, and included it in every figure. It is the reason the mean fee of €1.43 sits well above the median of €0.84. What lies behind the number we do not know. A data-maintenance error is conceivable, and so is a deliberately prohibitive setting for an asset that is not meant to be withdrawn right now. The entry's status was OK. We did not put the question to the provider and accordingly ascribe nothing to it.
There is a sound explanation for why an exchange's withdrawal fee can sit above the fee of a single transfer. An exchange usually bundles several withdrawal requests into one transaction and draws on a large number of existing balance fragments to do so. Such transactions are larger than the 141 virtual bytes of a simple case, and larger transactions cost more on the Bitcoin network. On top of that, an exchange cannot adjust a fee minute by minute and therefore has to price in a buffer for rising rates.
That explanation carries part of the measured gap. Whether it carries all of it cannot be judged from outside, because the size of the transactions actually sent and the internal calculation are not public. Our numbers establish the gap, not its cause. Anyone wanting to turn it into an accusation is going beyond what the data supports.
A second qualification concerns the direction of travel. The Bitcoin mempool was unusually cheap at the time of the survey. In periods of heavy load the relationship inverts, and the fixed withdrawal fee then sits below what your own transaction would have cost. If you can plan a withdrawal, look at the mempool rather than at an average figure from a guide.
The survey measures what a provider publicly states at a point in time. What is actually billed in the end, it does not measure. Only one of the several venues relevant to German investors is covered in full breadth, because only that one carries its fees in an open API; the cross-checks at KuCoin, Bitstamp and Coinbase Exchange concern individual assets and status entries, not the whole fee schedule. Bitpanda and Bison, which are closer to hand for many German investors, we could not measure, because their fee pages reject automated requests.
Also not covered are euro withdrawals to a bank account. A separate logic applies there, because no blockchain is involved. And the numbers say nothing about whether a provider is cheap overall: trading fees, spreads and custody terms can offset a low withdrawal fee or the reverse. A cost comparison between two platforms is rarely settled on the withdrawal alone.
(As of August 17, 2026. This article is not investment advice. Prices and fee structures change; check the terms with the provider before you buy.)
