Open a Bitcoin wallet and you see a number. It is easy to picture that number sitting in an account, the way a bank balance does. On the Bitcoin chain the number is a calculation. The wallet searches for unspent transaction outputs it can unlock, adds their amounts, and shows you the total.
An output is a coin, until it is spent
Every payment creates one or more outputs. Each output has an amount and a locking condition. If nobody has spent that output, it is an unspent transaction output, a UTXO. Spending it means writing a new transaction that names it as an input. Once that spend is confirmed, the old output is gone. It is not reduced. It is consumed.
Suppose three earlier payments left you outputs of 0.05, 0.20, and 0.015 BTC, and your wallet holds the keys for all three. The balance it shows is 0.265 BTC. There is no fourth object called the balance. If you pay 0.10 BTC, the wallet must pick one or more of those outputs as inputs. It cannot slice 0.10 out of the 0.20 output and leave 0.10 behind inside the same output. It spends the whole output and creates new ones, including change.
Why the same balance can produce different fees
A transaction that spends one output is smaller than a transaction that spends a dozen. Miners are paid, in part, by the fee, and the fee market is usually discussed per byte of transaction data. Two wallets can show the same balance and still build very different transactions, because one is holding a few large outputs and the other is holding many small ones. The small outputs are sometimes called dust when they are not worth the fee required to spend them. The chain still counts them. The practical ability to spend them is another question.
This is also why a wallet can say you have the money and then hesitate. It may be waiting for an output to confirm. It may be refusing to spend an output that is already used in an unconfirmed transaction. It may be hiding coins that sit on a script it does not yet know how to sign. The headline number is a sum under a set of assumptions.
What to ask the wallet
If the software lets you inspect coins, look at the list behind the total. Each row should point at a transaction output. The amount on that row is what will be consumed if you spend it. The address or script tells you why the wallet thinks the key in this backup can unlock it. A backup that does not cover every script the wallet used will restore a smaller sum, even though the chain has not changed.
The useful habit is to treat the balance as arithmetic you could repeat: find the unspent outputs, keep the ones this wallet can spend, add the amounts. Anything that cannot be tied to an output is not bitcoin on the chain. It is a label in an application.
Why two wallets can disagree about one backup
A seed can derive a long sequence of keys. Wallets do not scan forever. BIP 44 describes a gap limit: if the wallet sees a run of unused addresses, it stops looking further along that chain. A wallet that received funds on an address beyond another wallet’s gap will show a smaller balance until the scan is extended. The coins did not move. One program has not looked far enough.
Script type causes the same surprise. The same seed can back pay-to-script-hash addresses, native SegWit addresses, and Taproot addresses. A recovery tool that understands only one of those scripts will sum only the outputs locked to that script. The others remain spendable by software that knows the script, and invisible to software that does not. Before treating a restored balance as the whole story, ask which address types were scanned and how far the scan went. The chain’s answer is still the list of unspent outputs, not the first number a wallet was willing to display.
Sources
- Bitcoin white paper, on coins as chains of signatures.
- Bitcoin Developer Guide: wallets.
- Bitcoin Developer Guide: transactions.
- BIP 44, on account discovery and the gap limit.




