BitcoinKeys.guide

Back up your seed

Which metal backup should hold your seed?

“Move it to metal” is easy to say and oddly hard to act on — there are dozens of these things and they are not equally good. Here is how to choose one, which five we’d point you at, and which to avoid. We sell nothing and take no affiliate money, so nothing below is bought placement. New to this? The lesson on backing up your seed covers why metal at all, and how to copy the words without making a mistake.

Whose judgement this is

We rate hardware wallets ourselves, because that judgement is about design and firmware and we can read those. Metal backups are decided by fire, acid and a 20-ton press, and we have not run those tests. Jameson Lopp’s metal seed storage stress tests has, on 75 of them — the canonical independent testing in this space, and the source for every grade and price below. The criteria below are ours; every grade and price is his.

The good news first

Most of these products are fine. Of the 75 devices tested, 45 came through every test with nothing lost, another 11 were marked down in one test only, and just 10 — about 13% — did badly. So this is not a minefield, and you do not need to agonise. You need to avoid a short list of specific traps, and then pick on the things the stress tests don't measure. That second part is what the rest of this page is for.

What actually matters

  1. Stamped or centre-punched — not engraved, printed or etchedThe words have to be physically driven into the steel. Surface methods look fine on the shelf and disappear in a fire or an acid bath. This one distinction separates most of the passes from most of the failures.
  2. A single solid plate, not loose tilesTile systems ask you to slot dozens of little letter pieces into a frame. They work, but a crush or a drop can scatter them, and reassembling a seed from loose tiles is not something you want to discover under pressure. One plate has nothing to come apart. One tile design in the test set had its rivets dissolve, dropping every letter out — the tester’s own words for the result are catastrophic data loss.
  3. Stainless steel is enough; titanium is optionalGood stainless survives every test that matters — the cheapest device to score top marks throughout costs about the price of a takeaway. Titanium costs several times more and, in these tests, is not measurably tougher. Spend the difference on a second plate in a second location instead.
  4. You write it — no proprietary format, no company involvedA backup that needs a specific company’s product, app or lookup table to read back is a dependency you did not need. Plain BIP-39 words (or their numbers) that any wallet can accept is the whole point.
  5. Room for what you actually holdCheck it fits 24 words if you use 24, and that you have somewhere separate for a passphrase if you run one. A plate sized for 12 words is not a plate you can grow into.

Five that meet all five

Every one of these scored top marks in all 3 of Lopp’s stress tests and meets the 5 criteria above, which he doesn’t test for. Far more than five would qualify on durability alone — this is our shortlist, not a ranking, and deliberately not exhaustive.

Prices are bands rather than figures on purpose. We watch hardware-wallet prices on a schedule and stamp them with a date; we don’t watch these, so a precise number would be a promise we couldn’t keep.

The ones to avoid

The 10 that did badly are worth naming, and each failure shows you one of the criteria above doing its job.

Test results checked against the source, and every link above loaded, on 2026-08-06

✓ Last verified: August 6, 2026