X Shadowban Rate by Country, 2026: Hong Kong Leads at 55.5%
Which countries get shadowbanned most on X? 55.5% of accounts checked from Hong Kong had a restriction, vs 21.4% in Vietnam, across 6,466 accounts in 59 days.
Hong Kong has the highest shadowban rate on Twitter (X) in this study: 55.5% of the accounts checked by visitors from Hong Kong carried at least one active restriction, ahead of Japan at 50.0%. Vietnam is lowest at 21.4%. The world average across 6,466 accounts checked over 59 days is 42.9%. The gap between the highest and lowest country is more than 2 to 1.
What did this study measure?
This study measured how many X accounts had at least one of 4 restrictions active, grouped by the country of the visitor who ran the check.
Every time someone runs a free check on Shadowban Radar, the tool tests an X account against 4 signals and records the visitor's country.
The study covers every check run between August 1 and September 28, 2026. It keeps one record per X account (its most recent check in that window), for every country with at least 100 accounts checked.
Which 4 signals were tested?
The study tested 4 signals, independently:
- Search suggestion ban. The account no longer appears in the dropdown suggestions when someone types the handle or name in the search box, even though the profile and posts still exist.
- Search ban. The account's posts are missing from X search results entirely, even for people who already follow the account and search for exact wording.
- Ghost ban. Also called a thread ban or reply ban. The account's replies are detached from the conversation they belong to: everyone else sees the thread as usual, and the reply is simply not there.
- Reply deboosting. The account's replies are ranked below everyone else's, pushed under the sort order most people actually read a thread in.
Plain-language definitions of each term are in the X shadowban glossary.
What counts as a restricted account?
An account counts as "restricted" in this study if at least one of these 4 checks came back positive on its most recent check.
A signal that could not be measured on a given account is excluded from that specific signal's count rather than assumed either way.
Locked, temporarily limited and read-only accounts, which X announces itself, are a different case. They are covered in is my Twitter (X) account restricted?
Which countries have the highest shadowban rate on X?
Free, no login, results in seconds.
Hong Kong has the highest restriction rate in this sample, at 55.5%, and Vietnam the lowest, at 21.4%, among countries with at least 100 accounts checked.
| Country | Accounts checked | Restricted | Restriction rate | 95% confidence interval |
|---|---|---|---|---|
| Hong Kong | 292 | 162 | 55.5% | 49.7% to 61.1% |
| Japan | 1,212 | 606 | 50.0% | 47.2% to 52.8% |
| South Korea | 173 | 85 | 49.1% | 41.8% to 56.5% |
| Brazil | 161 | 78 | 48.4% | 40.9% to 56.1% |
| Saudi Arabia | 157 | 74 | 47.1% | 39.5% to 54.9% |
| United States | 846 | 396 | 46.8% | 43.5% to 50.2% |
| Indonesia | 276 | 116 | 42.0% | 36.4% to 47.9% |
| India | 247 | 92 | 37.2% | 31.5% to 43.4% |
| Nigeria | 1,065 | 379 | 35.6% | 32.8% to 38.5% |
| Thailand | 393 | 97 | 24.7% | 20.7% to 29.2% |
| Yemen | 110 | 27 | 24.5% | 17.5% to 33.4% |
| Vietnam | 159 | 34 | 21.4% | 15.7% to 28.4% |
World total: 42.9%, from 6,466 accounts (95% confidence interval: 41.7% to 44.1%). This counts all countries with a resolved location, including those below the 100-account cutoff.
For more figures, see X shadowban statistics, September 2026. The live figures are on the X shadowban statistics page.
How do you read the confidence intervals?
A 95% confidence interval is the range in which a country's true rate probably falls. It matters more than the single percentage, because it shows which gaps look real and which countries are effectively tied:
- A real gap: Hong Kong's interval (49.7% to 61.1%) does not overlap with Vietnam's (15.7% to 28.4%), so that gap looks real rather than noise.
- A tie near the top: South Korea, Brazil, Saudi Arabia and the United States overlap each other heavily, sitting in the same 47% to 49% band. Ranking them against one another beyond "roughly tied, well above the world average" would be reading more into the numbers than they support.
- A tie at the bottom: Yemen and Thailand have intervals that overlap almost entirely.
Where do Indonesia, India and Nigeria rank?
Indonesia (42.0%), India (37.2%) and Nigeria (35.6%) form the middle of the ranking, below the United States and above Thailand, Yemen and Vietnam.
Nigeria's sample is the second largest in this study, at 1,065 accounts. That gives it the tightest confidence interval of any country outside Japan (32.8% to 38.5%, a span of less than 6 points), so its position in the ranking is comparatively well established.
What can this study not show?
This study cannot show where the restricted accounts are based, or why the rate differs by country.
Does it show where the restricted accounts are based?
No. The country recorded here belongs to the visitor who ran the check, not to the X account being checked.
Someone in one country can check an account run from another country entirely. The reasons range from curiosity to customer support to following a public figure.
Read every row in the table above as "restriction rate among accounts that someone, checking from this country, cared enough to verify", not as a map of where X restricts accounts.
Does it explain why the rate differs by country?
No. Each of these is a plausible explanation, and none of them is tested here:
- Language
- The mix of account types that get checked in each country
- Local reporting or moderation patterns
- Plain differences in how much people in a given country distrust their own visibility
Treat any explanation you read for these numbers, including your own first guess, as an untested hypothesis until someone actually measures it.
The methodology page says more about how the 4 checks are defined.
Is 42.9% the real shadowban rate on X?
No. The 42.9% world figure, and every country breakdown built on top of it, describes people who were already worried, not the average X user.
Every account in this study was checked because someone used a free public tool. People mostly reach for a shadowban checker after they already suspect something is wrong, for example:
- A drop in replies
- A quiet feed
- A post that seems to go nowhere
That self-selection almost certainly pushes every number in this article above what a random sample of X accounts in each country would show.
How was the study run?
The study covers 6,466 X accounts, each counted once by its most recent check between August 1 and September 28, 2026. The method, in brief:
- The 59-day window is checked from a fixed pair of dates rather than a rolling "last 59 days", so the result does not shift with the day it is read.
- The country is the visitor's location at the moment of the check, resolved from IP address. Checks with no resolved country (about 1% of the total) are excluded from both the country table and the world total.
- Internal traffic (checks run by the people who built this tool, verifying their own accounts) is excluded throughout.
- The total differs from the September 2026 statistics, which cover other dates and apply their own exclusion rules.
- Countries below 100 accounts checked are left out of the table, since the confidence interval at that sample size gets too wide to be useful. This excluded Turkey, whose 97 accounts fell three short of the cutoff.
- Percentages are reported as a share, with 95% confidence intervals (Wilson score method, which stays well-behaved at these sample sizes rather than producing intervals that go below 0% or above 100%).
- No individual account, handle, or daily count is published, in line with how we report every study on this site: totals and percentages only.
To look closer, read X's own help page on locked or limited accounts and the code X publishes for its For You feed on GitHub.
What limits apply to every number?
Three limits apply to every number above:
- A snapshot, not a trend: every account is counted once, at its most recent check, so nothing here says whether a country's rate is rising or falling.
- A correlation, not a cause: nothing in this data says that X restricts accounts checked from one country more than another because of that country. It only shows that the restriction rate measured among people checking from there, in this window, was higher or lower.
- A self-selected sample: everyone behind these numbers chose to run a check, which is not the same as a random sample of X users anywhere.
Want to see where your own account stands on these 4 checks? Run a free shadowban check.
Choosing tools to grow an account? See best X growth tools.
Update history
- Corrected the study window to 59 days and the Turkey margin, and added how the total differs from the September statistics.
