Withdraw the same stablecoin on two different networks and the fee can differ by a factor of ten or more. Nothing about the coin changed. What changed is which network you asked to carry it, and networks are not priced alike.

Understanding the two pricing models makes the differences predictable rather than mysterious, and predictable matters, because the expensive mistake on this page is not overpaying by a few units. It is picking a network the other side cannot receive.

Why the same asset costs different amounts

The fee on the withdrawal screen is set by the exchange, as a flat charge per asset and network. What drives it is the network that moves the withdrawal: each one charges on its own terms, and the exchange's price follows those costs roughly rather than exactly.

Most assets exist on several networks. The balance on the exchange is just a number in their ledger; the network is chosen at the moment you withdraw. That choice is the entire cost decision.

Two typical fee models

Paying for computation

Some networks price each transaction by how much work it demands and how contested block space is at that moment, ethereum.org's gas documentation sets out the model, and a live gas tracker shows how far it swings inside a single day. Fees are therefore volatile — the same transfer can cost several times more during busy periods than during quiet ones, and the amount you are sending is irrelevant to the price.

That last part surprises people. Sending a tiny amount costs the same as sending a large one, which is what makes these networks unsuitable for moving small balances around.

Paying with resource allocation

Other networks give accounts an allowance that regenerates over time, TRON's resource model is the clearest written example. Transactions within the allowance are free or nearly free; beyond it, they are charged at market rates.

The practical effect is cheap and stable costs for ordinary users, which is a large part of why these networks dominate peer-to-peer settlement. The trade-off is a more complicated model — when something goes wrong, working out why is harder.

Choose in the right order: destination first, price second

This is the whole point of the page, so it gets stated bluntly.

Confirm the receiving side supports the network before you look at any fee

Funds sent on a network the destination does not support are, in the general case, unrecoverable. No fee saving is worth that outcome: the most you can save is a small fee, the most you can lose is the whole transfer.

The order to avoid is cheapest first, ask later. Picking a network because its fee is close to nothing, sending, and only then checking whether the destination supports it gets the sequence backwards. The saving would be real; so would the loss.

Trade-offs by network type, without the numbers

  • The oldest and most widely supported networks. Accepted almost everywhere, generally the slowest and among the more expensive. When compatibility matters more than cost, this is the safe answer.
  • General-purpose smart contract networks. Universally supported, fees that swing with congestion. Fine for larger amounts, poor for small ones.
  • High-throughput networks. Cheap and fast, very widely supported for stablecoins, which is why P2P settles on them.
  • Layer twos and newer chains. Often cheapest, but support is patchy. Verify acceptance explicitly rather than assuming. This is the category where the assumption fails.

Deliberately no figures here. Any number would be stale within weeks, and a stale number in this context is not a small inaccuracy — it is what makes someone choose wrongly.

The displayed fee is not the network fee

Exchanges usually charge a flat withdrawal fee per asset and network rather than passing through the exact on-chain cost. Sometimes that is more than the network charged, sometimes less. It is their pricing, not a measurement, which means comparing an exchange's displayed fee against a block explorer is comparing two different things. Compare their displayed fees against each other instead.

Consolidating is usually cheaper

With a flat fee per withdrawal, one transfer of a larger amount beats several small ones. Batch them where you can — but not so aggressively that you create a pattern of identical repeated transfers, which is its own problem on the receiving side.

Small leftover balances

Everyone accumulates them: a few units of something, change from a fee, a small airdrop. They share one property. The withdrawal fee exceeds their value, so they cannot be moved economically.

Options, in descending order of sensibility: convert them into something you actually hold using whatever consolidation feature the platform offers; leave them alone; or, the one to avoid, hunt for a cheaper network specifically to extract them. That last route is where people end up sending on a network the destination does not support — spending real money to rescue almost none.

The field next to the address

Some networks and some recipients require an additional identifier alongside the address. Memo, tag, or a similar name depending on the platform.

Leaving it out means the transfer succeeds on-chain but the recipient's system cannot tell whose funds arrived. The money reached the address; it is simply not attributed to you. Recovery depends entirely on the receiving platform's process, and some do not have one.

The reverse also happens: adding a memo where none was requested. Usually harmless, occasionally flagged as anomalous. The rule is narrow — supply one if you were given one, leave it blank if you were not, and never invent one.

Some transfers never touch a chain

Moving funds between accounts on the same platform is usually a ledger entry, not a blockchain transaction. Instant, and typically free.

Worth knowing because people withdraw and re-deposit to move funds between two accounts on the same exchange, paying a network fee to accomplish something the platform would have done internally for nothing.

Four checks before you press confirm

  1. Network. Does the receiving side accept this one? Confirmed, not assumed.
  2. Address. Paste it, never type it. Verify the first and last several characters against the source. The address format checker catches shape mismatches, though it cannot tell you an address belongs to who you think.
  3. Memo or tag. Required? Then filled in. Not required? Then empty.
  4. Amount. Above the destination's minimum, and you have accounted for the fee coming out.

If it has already gone wrong

Honest answers, which are mostly unwelcome:

  • Wrong network, address belongs to an exchange. Sometimes recoverable through support, often at a cost, never guaranteed. Ask early.
  • Wrong network, address is your own wallet. Occasionally recoverable if you control the keys and the networks are compatible enough. Technical, and easy to make worse by improvising.
  • Missing memo, sent to an exchange. Depends on their process. Open a ticket with the transaction hash immediately.
  • Wrong address entirely. If it belongs to a stranger, it is gone. There is no mechanism, and anyone who says otherwise is selling something.

Which is why the four checks above are worth the thirty seconds. How withdrawal fees sit alongside every other cost is in how fees stack.