Three confirmations cleared every ordinary fork our nodes recorded in 2026 on Bitcoin, Litecoin, Dogecoin and Dash. That's about 26 minutes on Bitcoin, 7 on Litecoin and Dash, and 3 on Dogecoin. Six buys you margin. Neither number would have covered the one event that stood out this year, a 13-block reorg on Litecoin in April.
We run our own full nodes for all four chains because our gateway reads the blockchain directly. A node remembers every competing branch it has seen, so we asked ours, and the answers are below. Data runs to 30 October 2026. You can download all 772 records as a CSV.
What the nodes recorded
A side branch is what's left over when two miners find a block at nearly the same moment. The network keeps one and drops the other. A reorg is the same thing seen from the losing side: blocks your node accepted get replaced.
| Chain | Period (2026) | Blocks covered | Side branches | 1 block | 2 blocks | Deeper | One every |
|---|---|---|---|---|---|---|---|
| Bitcoin | 9 Jan to 30 Oct | 42,314 | 18 | 17 | 1 | 0 | 2,351 blocks |
| Litecoin | 21 Jan to 30 Oct | 162,066 | 221 | 219 | 1 | 1 | 733 blocks |
| Dogecoin | 16 Feb to 30 Oct | 349,636 | 353 | 351 | 2 | 0 | 990 blocks |
| Dash | 27 Feb to 30 Oct | 134,384 | 176 | 176 | 0 | 0 | 764 blocks |
| Total | 688,400 | 768 | 763 | 4 | 1 |
More than 99% of side branches were one block long. Those are routine. A transaction in the losing block normally goes back in the queue and gets mined in the next one, and nobody notices. A reorg only costs a merchant money when someone uses it to spend the same coins twice.
The five that went deeper
| Date | Chain | Branch tip height | Depth | What we know |
|---|---|---|---|---|
| 21 March | Litecoin | 3,075,873 | 2 | Not traced |
| 23 March | Bitcoin | 941,882 | 2 | Written up on the BNOC forum as the reorg at height 941,880 |
| 30 March | Dogecoin | 6,145,010 | 2 | Not traced |
| 25 April | Litecoin | 3,095,943 | 13 | MWEB exploit, below |
| 9 September | Dogecoin | 6,367,679 | 2 | Not traced |
Our Litecoin node still holds that April branch: 13 blocks ending at height 3,095,943. Cryptonomist reported that attackers abused a flaw in Litecoin's MWEB privacy layer, that the reorg touched blocks 3,095,930 through 3,095,943 and took more than three hours to resolve, that double spends hit several cross-chain protocols, and that NEAR Intents estimated its loss at about $600,000. Bitbo covered it too.
To sit past that reorg you'd have needed 14 confirmations, about 35 minutes by our math. Few merchants wait that long on Litecoin, and I wouldn't change your settings because of it. This was a bug in a chain, not bad luck with miners. Confirmations protect you from bad luck. For a bug you need to watch the news and be ready to pause a coin.
What that means for your wait
| Chain | 3 confirmations, median (slow 10%) | 6 confirmations, median (slow 10%) | GriffNode default |
|---|---|---|---|
| Bitcoin | 26 min (53) | 55 min (91) | 6 |
| Litecoin | 7 min (13) | 14 min (23) | 6 |
| Dogecoin | 3 min (6) | 6 min (10) | 40 |
| Dash | 7 min (14) | 15 min (24) | 6 |
| Ethereum | Fork data not measured, see below | 20 | |
Times are from the 90 days to 30 October, measured block to block in our confirmation time dataset. They leave out however long a transaction sits in the queue before its first block, which depends on the fee the customer paid.
What follows is my judgment, not data. For a small digital sale you can switch off, one confirmation is fine: a stale block alone doesn't hurt you, someone has to actively cheat, and you'd lose one order. For normal physical orders, wait three. For anything you'd hate to lose, stay on six.
The fork tables measure accidents. They say nothing about someone paying to attack a chain, and that's the reason our defaults sit above three. A deliberate attack gets harder with every block, and how much harder depends on the chain's total mining power. Dogecoin's default of 40 is the most cautious of the lot, about 42 minutes by our math, and nothing in this year's fork data asks for it. If your Dogecoin orders are small, you can lower it in your wallet settings.
Two chains need a note. Dash has ChainLocks, built to protect against reorganisations, and every side branch we recorded there stopped at one block. Ethereum isn't in these tables because our Ethereum node is pruned and doesn't answer this query. It also works differently: ethereum.org describes finality as a supermajority link between two checkpoints, with a checkpoint every epoch of 32 slots at 12 seconds each. That puts finality at around 13 minutes, while our default of 20 confirmations takes about 4. For a large Ethereum order, give it another nine minutes before you ship.
How we measured this
Every Bitcoin-family node answers the same call:
bitcoin-cli getchaintips
Litecoin, Dogecoin and Dash have the same RPC. It returns each branch tip the node knows about with a branchlen, which is the depth in our tables, and a status. The Bitcoin Core reference defines them. We counted these:
| Status | Definition | Records |
|---|---|---|
| valid-fork | "This branch is not part of the active chain, but is fully validated" | 294 |
| valid-headers | "All blocks are available for this branch, but they were never fully validated" | 365 |
| invalid | "This branch contains at least one invalid block" | 106, all Dash |
| conflicting | Dash only, not in the Bitcoin reference | 3 |
We left out four records with the status headers-only, meaning "Not all blocks for this branch are available, but the headers are valid". Our node never saw those blocks, so we can't vouch for them. One of the four is a 2-block Bitcoin tip at height 961,633 on 8 August. It's in the CSV, flagged.
Read these numbers with their limits in mind:
- It's a lower bound. A side block that never reached our nodes isn't here, and a different node would show a slightly different list.
- The periods differ. Each one starts at the first side branch that node has on record, which is why Bitcoin starts in January and Dash in late February.
- Litecoin's March is odd. We recorded 75 side branches that month against about 20 in a normal one. We don't know why.
- Timestamps come from miners. Dates here can be off by minutes.
If you sell on Shopify, our Shopify guide shows where this wait fits into the order flow.
Frequently asked questions
How many confirmations does a Bitcoin payment need?
For an ordinary order, three cleared every fork our Bitcoin node recorded in 2026, at a median wait of about 26 minutes. For a large order keep six, about 55 minutes. A small digital sale you can switch off is fine on one.
How long do three confirmations take?
Measured over the 90 days to 30 October 2026, the median was about 26 minutes on Bitcoin, 7 on Litecoin, 7 on Dash and 3 on Dogecoin. The slowest 10% took more than 53, 13, 14 and 6 minutes.
What is a blockchain reorg?
A reorg is the network replacing recent blocks with a competing branch. Most are one block long and harmless, because transactions from the replaced block usually get mined again. Money is at risk only when someone deliberately spends the same coins twice.
Emil Dabrowski is the founder of GriffNode, a non-custodial crypto payment gateway based in Poland. GriffNode sets the default confirmation counts quoted here, so weigh this page with that in mind. More about Emil and the company.