How to mute words on X (Twitter)

X buries muted words in a different spot on each platform, which is why so many people give up looking. Here is the current path on all three, as of June 2026.

Four facts decide whether your mutes actually work, and they are the same on every device. (1) Muting is case-insensitive: add CATS and you also mute cats. (2) Muting a word also mutes its hashtag, so unicorn covers #unicorn too. (3) You can mute a full phrase, an emoji, or @handle to kill mentions of an account without muting the account. (4) Mutes apply to your timeline and notifications, but not to search. You hid the word, you did not delete it.

The add-a-word screen has three settings people click past. Each one changes what the mute actually does, and the right combination depends on why you are muting.

• Home timeline vs Notifications. Turn on both for a topic you never want to see. Turn on Notifications only when you still want the word in your feed but are tired of the pings (a brand handle you monitor, say). Turn on Home timeline only for spoilers you would rather not stumble into mid-scroll.
• From anyone vs From people you don't follow. This one applies to notifications only. "From people you don't follow" is the setting that quietly fixes most reply spam: it silences strangers using the muted phrase while letting the people you actually chose still reach you.
• Mute timing. Forever is the default, but 24 hours, 7 days, and 30 days exist for a reason. A finale spoiler needs a week, not a life sentence. A trending hashtag needs a day. Use the timer and your muted-words list stays short instead of becoming a graveyard of dead topics.

One honest limit before you build a long list: X has no wildcards. Muting spoiler is not guaranteed to catch spoilers, and it will not catch a scammer writing airdr0p with a zero. Add the plural and the obvious misspellings as separate entries. It is tedious, because X only lets you add words one at a time, which is also the real answer to "is there a limit?" There is no published cap on how many words you can mute; the friction is the one-by-one entry, not a ceiling. To undo one, open the same Muted words list, tap the entry, and delete it. That is also where you go to change its duration or scope later.

Most "words to mute" lists are one person's political feed turned into a screenshot. Useful for them, not for you, and trapped in an image a search engine cannot read. Below is a plain-text, categorized list you can actually copy. It splits into two jobs: cleaning up your own feed, and cutting the spam and bot noise that follows every account once it grows. We build a bot-filtering product, so the second half is where we spend our time.

The spam categories are the ones that pay off most, because they are evergreen and non-partisan. Nobody's feed is improved by a "first 100 to reply gets the airdrop" post. If you want the reasoning behind those scam tells, the same patterns drive how to spot a Twitter bot. And before you mute every engagement-bait phrase out of spite, know that the same tactics quietly tank the poster's reach too, which is half of what the algorithm actually rewards.

Muting a word is one of four tools that look similar and behave differently. The line that matters: mute is invisible, block is not. A muted account never knows. A blocked account finds out the second it lands on your profile. Here is the full comparison.

There is a fifth control worth knowing, sitting one screen over in Settings → Notifications → Advanced filters. It lets you silence notifications from accounts that you don't follow and that are brand new, have a default profile photo, or have never confirmed an email or phone number. Every one of those toggles is scoped to people you don't follow, so it never touches your real circle. It is the closest thing X gives you to a built-in bot filter, and that is exactly where the next section comes in.

The top complaint about muted words is that the word keeps showing up anyway. Five real reasons explain almost every case, and only one of them is a bug.

1. You follow the account posting it. If you opted into mobile notifications for someone you follow, X still pushes their posts to you even when they contain a muted word. The mute loses to the follow.
2. You found it in search. Mutes never apply to search results. This is by design, not a glitch.
3. It is a near-miss. No wildcards means crypto does not catch crypt0, and spoiler may not catch spoilers. Scammers and shitposters lean on this on purpose.
4. The For You feed re-surfaces the topic. The algorithmic timeline recommends posts based on engagement, and a muted word does not always stop a recommended post from a stranger. Switching to the Following tab makes mutes feel far more reliable.
5. It is in a quote or reply you do see. Mutes cover replies and quote posts of muted terms, but a followed account quoting the topic in their own words can still slip through.

Every few months a list of internal X tags makes the rounds (ActivityTweet, suggest_recycled_tweet, suggest_who_to_follow and a dozen cousins) with the promise that muting them kills "In case you missed it," recycled posts, and follow suggestions. It is a tempting trick. It also does not work.

The honest version: the most-shared copy of that list has a top comment, left on the gist itself, that reads "this mute list does not work, and has never worked." X's own support guidance says muting those strings will not reduce suggested posts. They were never user-facing labels you could filter against. Paste them in if you like, but spend the effort on the categories above instead. The thing that actually quiets recommendations is using the Following feed, not muting a tag that was never exposed to you in the first place.

Muting is a filter for what you consume. It is the right call for spoilers, outrage, and the spam in your own replies. It is the wrong call the moment your goal flips from "see less" to "catch everything that matters and skip the junk." You cannot mute your way to monitoring a brand, because muting hides posts, and monitoring exists to surface them.

That is the split we built ReplySocial around. A muted word filters one feed by one string. A saved keyword monitor does the opposite job: it watches X, Reddit, Facebook, and LinkedIn for the words youwant, drops the matches into one inbox, and uses bot scoring to keep the spam out of that queue automatically. The applied version is in how to monitor brand mentions, and the full stack across networks is in the social media monitoring guide. If you would rather just hand-build the search instead, our free X search query builder and the advanced search operators reference cover the manual route.