How does the mechanism work?
A confirmation count describes a transaction’s position in the chain a node currently accepts. The work in that history matters when considering competing chains, but a count is not a certificate from a central operator.
Nodes can temporarily disagree about the latest block. When the valid chain with the most accumulated work changes, a reorganisation can replace blocks. A transaction in a replaced block may later be included again, remain unconfirmed or conflict with the accepted history.
Always attach the count to an observer’s chain view. Two interfaces can temporarily show different tips or update at different moments without changing the arithmetic. A custodial account can add another delay before displaying a credit. Ask which block contains the transaction and which tip is being used before comparing counts. If inclusion itself changes during a reorganisation, recalculate from the accepted history rather than adding elapsed time to an old count. This approach also keeps confirmations separate from authorisation: a later block does not repair a bad signature, change the recipient or ask the sender to approve the transaction again.
What would you see in an example?
Suppose a wallet first shows a payment as pending. A block then includes it, giving it one confirmation. A later block builds on that block, giving it two. The second block need not mention the payment directly; the count grows because the history containing it has gained a descendant.
How do you calculate a count from two heights?
For a transaction included at height H in the chain you accept, and a current tip at height T, the confirmation count is T - H + 1. The plus one counts the containing block itself. An unconfirmed transaction does not have a containing height in that accepted chain, so the same subtraction is not applied to a pending payment.
Use an invented example: the containing block has height 900,000 and the tip has height 900,002. The count is 900,002 - 900,000 + 1 = 3 confirmations. Only the first of those blocks contains the transaction. The later blocks commit to prior history through their previous-block hashes. They do not need to repeat its bytes or collect a fresh signature from its sender.
Now imagine a competing valid branch replaces the block that contained it. Your old subtraction no longer describes inclusion in the accepted chain. If the transaction is included again at a different height, calculate the count from its new containing block. If it is absent, it may become unconfirmed; if a conflicting spend is accepted, its inputs may no longer be available.
A reorganisation requires comparison of valid histories and their accumulated work, not a vote by a fixed number of peers. A node checks the replacement history under its rules. The confirmation count is a useful summary of position, but it omits difficulty, competing work, transaction dependencies and network conditions. Those omissions explain why a count alone cannot guarantee finality for every situation.
Common misunderstandings
Confirmations do not measure elapsed minutes. Several blocks can arrive close together or take longer than an average-spacing estimate. A clock and a chain-height difference answer different questions.
An account provider’s credited balance may follow its own policy. That display can lag inclusion or apply additional conditions, while an accepted transaction is already visible to a validating node. Keep the provider’s accounting state separate from the protocol count. This explanation supplies no universal number at which a payment is safe.
Check yourself
A payment is in the current tip block. Does it have zero or one confirmation?
Reveal the answer
One. The containing block counts. Zero normally describes a transaction that has not been included in the accepted chain.
What does this diagram leave out?
There is no universal count that makes every transaction risk-free. The white paper explains confirmation reasoning under a model with assumptions; it is not a promise about every network condition. This page gives no settlement recommendation. Always distinguish the chain view, the transaction status and a custodial service’s own crediting policy.
Which primary sources explain the rules?
Sources: Bitcoin white paper; Block chain developer guide.
Sources: Confirmation count semantics.