Block Reorganization (Reorg)

A block reorganization, or reorg, happens when a node abandons one or more blocks at the tip of its chain and adopts a competing branch that carries more accumulated proof-of-work. Bitcoin nodes always follow the most-work chain, so when two miners find blocks at nearly the same height a temporary fork forms; the branch that next extends fastest wins, and the losing blocks are discarded.

How reorgs occur

Block propagation across the global network is not instant, so brief competing tips are normal. Most reorgs are one block deep and resolve within minutes when the next block extends one branch. Transactions in the orphaned block return to the mempool to be re-mined, unless they were already included in the winning branch.

Why depth matters

Shallow reorgs are routine and harmless to confirmed transactions, which is precisely why merchants wait for several confirmations. A deep reorg, one that rewrites many blocks, would require an attacker to out-mine the honest network over a sustained period, the same capability behind a 51% attack. No such deep reorg has ever succeeded on Bitcoin's mainnet, though smaller chains have suffered them.

Reorgs are the mechanism behind Bitcoin's probabilistic finality: a transaction is only ever as final as the work piled on top of it is hard to redo. Blocks discarded in a reorg become orphan or stale blocks, and a reorg deep enough to undo a payment is the practical aim of a double spend.

What Happens to Transactions in a Reorg

When a block is orphaned, its transactions do not vanish — they return to the node's mempool as unconfirmed and are almost always re-mined within a block or two, since the winning branch's miners want the same fees. The important exception is the coinbase reward itself: it exists only in the discarded block and has no counterpart to return to. This is exactly why consensus rules force newly mined coins to mature for 100 blocks before they can be spent — a rule that shields the whole economy from spends whose funding could be erased by an ordinary reorg. A payment that was in both competing branches is unaffected; only its confirmation count shifts.

The Miner's View: Stale Work and Wasted Blocks

For miners, every reorg means someone did real work that earned nothing. The block that loses the race becomes stale, and its finder forfeits the full subsidy plus fees — a strong economic incentive behind the industry's obsession with block propagation speed, compact relay, and low-latency pool infrastructure. Pools react to a new chain tip by immediately pushing fresh job templates; shares submitted against the old tip in that window are stale shares, which is one of the quiet ways poor connectivity taxes a mining operation even when the hardware is perfect.

How Often, and How Deep

On modern Bitcoin, small reorgs are rare and shallow — typically a single block, resolved in minutes — because propagation has improved enormously since the network's early years. The notable deep reorganizations in Bitcoin's history came not from attackers but from software faults, most famously the March 2013 database-incompatibility fork between node versions, which the community resolved by deliberately converging on one branch. That history is why confirmation-depth guidance exists on a sliding scale: one confirmation for coffee, several for meaningful sums, and more when accepting payment from parties with both the means and the motive to attempt a double spend. For a node operator, watching your own node log a reorg is worth doing once: it makes concrete that the chain tip is a live negotiation, not a fixed record, until depth settles the question.