The Government Crypto Holdings database records how much cryptocurrency national governments control, where it came from, and what happened to it — with a source for every figure. This page explains the rules behind the table.

What counts

An entry covers a national government or a state-controlled vehicle: law enforcement custody, treasuries, central banks, and sovereign funds (e.g. Bhutan's Druk Holding). Each row records:

  • Origin — how the crypto was acquired: seizure, purchase, mining, donation, or theft (state-sponsored hacking; North Korea's Lazarus Group is tracked as stolen holdings, not a reserve).
  • Status — holding, partially sold, sold, or disputed.
  • Confidence — confirmed (official statements or well-corroborated on-chain attribution), estimated (attribution with disclosed uncertainty; shown with ~), or disputed (evidence contradicts the claim).
  • A timeline of dated events — seizures, court rulings, sales, laws — each with a link where a stable source exists.

What the headline total counts

The BTC documented figure at the top of the page counts bitcoin a government holds as its own. The total is prefixed with ~ whenever any included row is estimated. It is a documented estimate, not an audit of government wallets. Three things are deliberately left out of it, while staying in the table:

  • Disputed claims. Bulgaria's famous 213,519 BTC and China's PlusToken 190,000 BTC are in the database because the claims are worth documenting — and out of the total because the evidence contradicts them.
  • Stolen funds. North Korea's Lazarus holdings are proceeds of exchange and bridge hacks, actively being laundered, with a balance that changes weekly. Folding them into the same number as the U.S. Strategic Bitcoin Reserve would make the total mean less, not more.
  • Non-BTC assets. The total is denominated in BTC, so the ETH rows don't contribute to it. They are listed, with their own amounts, in the table.

The table itself ranks BTC rows first and orders by amount within an asset. Amounts in different assets are not comparable as numbers, and this dataset carries no price feed to rank them by value — so it doesn't pretend to.

Seized is not owned

Coins in law-enforcement custody are not state property until forfeiture is final. The U.S. holding includes coins pending forfeiture — the Chen Zhi coins are in litigation and the Bitfinex recovery is under court-ordered restitution to the exchange; the U.K.'s 61,245 BTC are subject to an ongoing civil recovery in which victims claim most of the sum; Finland's remaining 90 BTC await a final judgment.

Stating that rule and then publishing one combined number was the wrong way round, because the combined number is overwhelmingly custody: a reader quoting "BTC documented" as government-owned would be wrong about roughly 96% of it. So the page now carries two figures. BTC documented is everything a government controls and counts toward the dataset. BTC owned outright is the part acquired by purchase, mining or donation — coins whose title nobody is litigating. The difference is seized stock, which stays in the total and in the table because governments do control it, but is not the same claim.

Which one to cite depends on the sentence. "Governments control X bitcoin" is the first. "Governments own X bitcoin" is the second.

How amounts are verified

Governments almost never publish wallet addresses (El Salvador is the main exception), so most figures rest on on-chain attribution by analytics firms — primarily Arkham Intelligence — corroborated by court documents and official statements. Attribution is probabilistic: where the perimeter is uncertain (e.g. the UAE's state-vs-royal-family wallets), the entry is marked estimated.

Transfers to exchange deposit addresses are treated as presumed sales when no official confirmation exists — the pattern seen in Germany's 2024 liquidation and Bhutan's 2025–26 sell-down.

Why some famous figures are marked disputed

Aggregator sites copy each other, and several widely-cited numbers do not survive contact with primary sources:

  • Bulgaria's 213,519 BTC — a 2017 press extrapolation. Prosecutors stated in 2022 that no such coins were ever confiscated or held.
  • Ukraine's 46,351 BTC — aggregated personal asset declarations of public officials, mislabeled by trackers as a state treasury. Ukraine has no crypto reserve; the enabling bill has not passed.
  • China's ~190,000 BTC — ruled forfeited to the treasury in 2020, but analysts believe the coins were sold in 2019, and no evidence of current custody exists.

Disputed rows stay in the table as documented debunks, render muted, and are excluded from the headline totals.

Review cadence

Entries are reviewed monthly against sources, and immediately after major events (court rulings, executive orders, large attributed movements). Every row carries its own "as of" date — figures are only as fresh as that date, and dates keep the precision the source supports (a year, a month, or a day).

Change log

The total in a published citation should stay checkable after the records move on. Every change to the data or its rules gets a dated entry here:

  • 2026-07-25 — the page now publishes BTC owned outright alongside BTC documented, applying the "seized is not owned" rule to the numbers instead of only to the prose. Bhutan corrected from 3,954 to 3,119 BTC and downgraded from confirmed to estimated: Druk Holding says it does not recall selling any bitcoin, so the balance is on-chain attribution the holder contradicts. The U.S. row keeps 328,372 BTC but is re-dated to 2026-02-17, the vintage of the Arkham attribution it rests on — the earlier 2026-07-16 date implied a freshness the source could not carry — and its citation moved off an aggregator to the report of that attribution.
  • 2026-07-17 — holdings that originate from thefts (e.g. state-linked hack proceeds) and disputed rows no longer count toward the headline BTC total; they stay in the table, marked. Table now sorted by asset.
  • 2026-07-16 — initial release: 12 countries, per-row sources and as-of dates.

Corrections

Every entry is public and attributable. If we got a fact wrong, tell us — Press & contact — and the correction ships with the next deploy.