Chain Reorganization (Reorg)

Chain reorganization, or reorg, is what happens when a node abandons blocks it had previously accepted at the chain tip and switches to a competing branch that carries more cumulative proof of work. Bitcoin nodes always follow the chain with the greatest total work, the principle described under the longest-chain rule and, more precisely, the heaviest-chain rule, so when a heavier branch appears, the node rewinds its recent history and replays the winning side. Nothing about this is an error condition; it is the mechanism by which a network with no coordinator converges on a single history.

The possibility of reorganization follows directly from Bitcoin's core design choice: there is no authority to declare which history is official, so nodes must be prepared to change their minds when the evidence changes. Accumulated proof of work is the evidence, and a reorg is simply a node updating its belief to match the greatest accumulation. What keeps this from being chaos is economics rather than decree. Producing a heavier competing branch costs real energy, and the deeper the target, the more energy an attacker must burn while the honest majority keeps extending the public chain, so honest convergence is overwhelmingly the cheapest outcome for everyone. Understanding reorgs is understanding why Bitcoin speaks of settlement in probabilities: a payment is never final by proclamation, only buried beyond any rational attacker's budget.

Routine reorgs

Most reorgs are shallow, accidental, and harmless. When two miners find valid blocks at nearly the same height, propagation delay leaves parts of the network temporarily following different tips, a race explored further under block propagation. Whichever branch the next block extends becomes the heavier chain, and nodes on the losing side reorganize to match. The displaced blocks become stale blocks, their miners lose the reward, and any transactions unique to them return to the mempool to be mined again; coins from a staled block's coinbase can never circulate, which is one reason coinbase maturity forces miners to wait 100 blocks before spending rewards. A one-block reorg of this kind is a normal, expected event and is usually invisible to users.

When reorgs become weapons

Deeper, deliberate reorgs are the mechanism behind Bitcoin's most serious theoretical attacks. An entity commanding enough hashrate can mine a competing branch in private and release it once it is heavier, erasing confirmed transactions from the public history, which is the core maneuver of a 51% attack and the industrial-scale form of a double-spend. Smaller-scale variants like the Finney attack and race attack exploit the same window at shallower depth. The defense is arithmetic: each additional block a transaction is buried under multiplies the work an attacker must redo while the honest network pulls further ahead, so the probability of reversal falls off steeply with confirmation depth. Six confirmations remains the long-standing convention for high-value settlement, less a magic number than a comfortable margin.

What operators actually see

For a node operator, a reorg appears in the logs as a disconnect-and-reconnect of tip blocks; for a miner, it occasionally appears as a block reward that evaporates. Both are the system working as designed. Bitcoin does not offer instant finality and never claimed to; it offers probabilistic finality that hardens by the minute and, a few blocks deep, becomes as close to irreversible as anything in engineering. Reorgs are the visible seam of that process, the place where competing histories are resolved by accumulated work rather than by authority. A network that reorganizes shallowly and rarely is not fragile. It is a network honestly reporting how consensus is made. That honesty is the feature, not the flaw.