Twitter (X) Direct Message Limit: the Daily DM Cap, Who Can Message You, and Why DMs Fail to Send

The table lists each limit, who it applies to, the number if X publishes one, and where it comes from. "Not published" means we could not find the number on any X page or post we read; figures that circulate on third-party sites without a source are left out.

The 500-a-day figure. X's help center page on limits (help.x.com/en/rules-and-policies/x-limits) lists "Direct Messages (daily): The limit is 500 messages sent per day." The same page says the limits include actions from all devices, including web, mobile, phone and API, and that they may be temporarily reduced during periods of heavy site usage, in which case X posts an update on its status site. The help center blocks automated fetching, so we confirmed the wording through two independent reproductions of the page, including The Register's May 2026 article on X's tightened free-account limits, which quotes the same line.

What the page does not say. It gives no separate DM number for unverified accounts, no number for Premium tiers, and no character limit for a single message. Third-party guides fill these gaps with figures that do not trace back to X; we leave them out rather than repeat them.

Do not confuse it with the post limits. The same help page caps unverified accounts at 50 original posts and 200 replies a day, and follows at 400 a day. Those are separate counters; sending DMs does not use up your posts, and the reverse. If your problem is a post that will not publish, see our page on why a tweet fails to send (/blog/tweet-failed-to-send).

Verified accounts (Premium, Business, Government): up to the 500-per-day ceiling, as far as X has published. X has never said that paid tiers get more than 500, and we found no X post that does.

Unverified accounts: less. The change was announced by X's support account in July 2023, quoted by Fortune at the time: "We'll soon be implementing some changes in our effort to reduce spam in Direct Messages. Unverified accounts will have daily limits on the number of DMs they can send. Subscribe today to send more messages." No number was given then, and none has been published since. In practice the limit is a moving target that X adjusts as part of its spam work.

The other half of the verification gap is reach, not count. In May 2026, Nikita Bier, X's head of product, replied to an unverified user: "You can't even DM half of the site without verification." That matches what message-request settings allow: a recipient can accept requests from verified accounts only, and an unverified sender cannot reach them at all, no matter how few messages it has sent that day.

When does the DM limit reset? The help center says that for time-based limits such as DMs you can try again after the time limit has elapsed, and that X tells you with an error which limit you hit. It does not say whether the window is a rolling 24 hours from your first message or a fixed daily reset. If you hit it, the safe assumption is a full 24 hours from the burst that triggered it; sending more attempts in the meantime does not help and can look like the spam the limit exists to stop.

Replies inside an existing conversation count too. The limit is on messages sent, not on new conversations. A long back-and-forth with one person uses the same allowance as 500 first messages to 500 people, but the second pattern is the kind of DM spam X says its limits exist to curb.

Every X account decides who may start a conversation with it. People you follow can always message you. For everyone else the recipient's settings decide; X's safety account pointed in 2023 to Settings, Privacy and safety, Direct Messages and the option "Allow message requests from everyone", and the inbox can also be limited to verified accounts. Since September 2023 creators can allow DMs only from their subscribers, which X announced as "creators now have the option to always (or only) allow DMs from subscribers". X's head of product suggested exactly that to a large account in 2025 as a defense against DM spam.

Message requests. A message from someone you do not follow does not land in your main inbox. It goes to Message Requests, where you accept, delete or report it. A message that was delivered but never answered usually sits there, which is why "I sent it but they never saw it" is so common. X's head of product said in 2026 that a redesigned Message Requests view was in testing.

Blocks and DM blocks. If someone blocked you, you cannot message them. Since September 2024 X also has a DM-only block (X's announcement: "DM block now live, for when you want to still hear someone's thoughts, just not directly"), so you can be able to see and reply to someone's posts and still be unable to DM them.

canDm is how this shows up in data. X's profile data carries a boolean that says whether the profile accepts DMs from accounts it does not follow. Our API returns it as canDm on /twitter/user/info and /twitter/user/batch_info_by_ids. X does not document the field, and it may be computed relative to the account doing the lookup rather than to you, so treat it as "open to strangers" and still expect your own send to fail if the account allows only verified senders and you are unverified.

We looked up 1,042 public profiles with at least 10,000 followers (median 64,899) that twelve organization accounts follow: @NASA, @nytimes, @Reuters, @BBCWorld, @NBA, @NFL, @premierleague, @UN, @Google, @Microsoft, @espn and @Spotify. These are the accounts people most often try to reach by DM: brands, teams, agencies, newsrooms, journalists and athletes. 432 of them, 41.5%, had canDm true.

Size matters more than the badge. Openness falls steadily with audience: 47.0% of accounts with 10,000 to 100,000 followers accept DMs, 39.9% at 100,000 to 1 million, 24.8% at 1 to 10 million and 14.8% (4 of 27) above 10 million. Within blue-check accounts the drop is sharper still: 52.0% open at 10,000 to 100,000 followers, 22.0% at 1 to 10 million.

Government accounts are the most closed. Only 17.2% of gray-check accounts accept DMs, and 4.2% of those with 1 to 10 million followers do. Agencies route contact through forms and press offices, and their DM settings reflect it. Accounts that carry an affiliate badge (a small logo of the organization they belong to) were open 28.7% of the time, against 43.8% without one.

Only 26 of the 1,042 profiles were protected and none carried the automated-account label, so the closed inboxes are a choice, not an artifact of private accounts.

Split by the organization that follows them, the communities differ more than the tiers do. Profiles followed by @nytimes, mostly its reporters and editors, were open 72.6% of the time: journalists tend to want tips. Those followed by @Reuters were at 52.6%. At the other end, the agencies and missions followed by @UN (18.9%) and @NASA (21.1%) mostly do not take messages from strangers.

What this means for you. If the account you want to reach is a government body or has millions of followers, expect a closed inbox and use its published contact route. If it is a reporter, a mid-size creator or a team account, a DM has roughly even odds of being accepted as a request, and a short, specific first message is what decides whether it gets read.

"Message failed to send", "Message not sent" and a red exclamation mark next to a DM are the same family of failure. Work through the causes in this order; the first three explain most cases.

1. The recipient does not accept your message. They allow requests only from people they follow, only from verified accounts, or only from subscribers. Fix: none from your side, other than following the account and waiting for a follow back, or getting verified if their setting is verified-only. Check canDm on their profile before you write a long message.

2. You are blocked or DM-blocked. A full block or the DM-only block stops delivery. Fix: none; respect it and do not try from a second account.

3. You hit the daily limit. The 500 ceiling, or the lower unpublished one on unverified accounts. X says the error names the limit. Fix: stop and wait until the window has passed; repeated retries extend the problem.

4. The app is out of date or mid-migration. X moved DMs to the new encrypted XChat in late 2025 and asked users to update their apps to see Chat and legacy DMs in one inbox; an old app build can fail to send into converted conversations. Fix: update the X app or use the standalone XChat app, then try again.

5. Your account is limited. If X has temporarily restricted the account (a verification prompt, a locked-account screen, a spam warning), DMs are among the first features to fail. Fix: complete the verification X asks for, then send slowly. Our page on the "request looks like it might be automated" error (/blog/twitter-request-looks-like-it-might-be-automated-error-226) covers that screen.

6. X has lowered limits or is having trouble. The help center says limits may be temporarily reduced during heavy site usage and that X posts an update on its status site when it does. Fix: check the status site and try later.

7. Group chat full. Adding someone to a group fails once it reaches the member limit (see the next section).

X replaced its old direct messages with XChat, an encrypted messaging system with file transfers and audio and video calls. Elon Musk described it in November 2025 as "our new communications system that has replaced the old Twitter DMs". X announced the rollout on iOS and web the same month ("update your X app to access Chat and your legacy DMs in one unified inbox"), launched a standalone XChat app on iOS in April 2026 and on Android in June 2026. Nikita Bier said the migration to encrypted DMs took three months.

Group chats grew. On April 22, 2026, X announced joinable links for group chats "with support for 350 members per chat (and growing)" and said it would deprecate X Communities on May 6. A day later Bier wrote that group chat limits would rise to 500 members "tomorrow" and that X aimed to reach 1,000 within weeks. We did not find a later post confirming 1,000, so 500 is the last confirmed figure.

What did not change. None of X's XChat announcements we read mentions a new daily send cap, so the 500-per-day figure on the help center is still the published ceiling. X's head of product said in October 2025 that after purging reply-spam bots X would be "focusing on DM spam next"; tighter enforcement is more likely to show up as accounts being limited than as a new published number.

For developers. X's developer account announced in March 2026 that the X Activity API delivers events for XChat DMs. The official DM endpoints and their limits are covered below.

If you send DMs from code, a second set of limits applies on top of the account's own. On the official X API (docs.x.com/x-api/fundamentals/rate-limits), each of the three send endpoints allows 15 requests per 15 minutes and 1,440 per 24 hours per user, and the app as a whole 1,440 per 24 hours. DM event lookups are 15 per 15 minutes. On the pay-per-usage rate card (docs.x.com/x-api/getting-started/pricing), a DM interaction create costs $0.015 per request and a DM event read $0.010 per resource. These limits do not raise the 500-a-day account limit; the help center counts API sends against it too.

For reads the gap is the familiar one: $0.005 per post read on X's card is $5 per 1,000 tweets, against $0.15 per 1,000 on twitterapi.io, about 33x, and the profile lookups above are about 55x apart. Pay-per-usage on X is capped at 3 million post reads a month.

Sending through twitterapi.io. send_dm_to_user exists for replying to people who messaged you or for one-to-one conversations you would otherwise type by hand. You log in once with /twitter/user_login_v2 ($0.10, with a proxy you supply) to get login_cookies, then each send is $0.0006. The call acts as that account, so X can rate-limit, lock or suspend it; keep a human pace; the session can need a re-login; and every call that returns HTTP 200 is billed, including ones whose body says status: "error". The endpoint's own documentation says you can only send DMs to people who have enabled them. It is not a way around the 500-a-day limit, and it is not for messaging lists of people who never asked to hear from you.

The useful part is the read side. Before anyone writes to a list of partners, reporters or customers by hand, a canDm lookup tells them which inboxes are open. The script at the end of this page audits the accounts any handle follows and prints the open share by verification type, without sending anything.

Question. What share of public X accounts accept direct messages from accounts they do not follow, and how does it change with verification type and audience size?

Accounts. 1,042 public profiles with at least 10,000 followers, taken from the accounts followed by twelve organization accounts: @NASA, @nytimes, @Reuters, @BBCWorld, @NBA, @NFL, @premierleague, @UN, @Google, @Microsoft, @espn and @Spotify (up to 95 per organization; @BBCWorld follows only 17 qualifying accounts, @premierleague 75). Each profile is counted once, under the first organization it appeared for.

Method. GET /twitter/user/followings supplied the ids only, because that list's own DM field is a placeholder; every profile was then read with GET /twitter/user/batch_info_by_ids, 100 ids per request, for canDm, verifiedType, isBlueVerified, followers, protected, isAutomated and createdAt. Business and Government come from verifiedType; "Blue check only" is isBlueVerified without a type.

Window and size. A single snapshot, collected in 2026: 34 requests, 1,042 profiles.

What the sample cannot show. It is not a random sample of X: it covers accounts large organizations follow, which skew toward media, sports, tech and government. canDm reflects the account's setting as seen by the account doing the lookup, so a verified-only inbox can read as open to one viewer and closed to another. It says nothing about how many DMs any account sends, and X's daily DM limits cannot be measured from public data.

Privacy. Aggregates only. No individual profile is named here except the twelve organizations used as seeds; raw lookups were not published, and the de-identified rows keep only verification type, follower bucket, account-age bucket and the DM flag.