Skip to content

GuideLimits

The X DM Limit Study: when the cap resets, and what X does at it

X DM limit reset, measured on 30 accounts over 21 days: when the next cold DM went through, what X showed at the cap, why retries failed, plus a CSV.

X's cap on cold DMs does not reset at midnight. Across 30 X accounts we tracked for 21 days, the next first message went through 24.7 to 27 hours after each batch began. X refused 1,078 sends along the way and suspended none of the accounts.

We call this the X DM Limit Study. We make the software that sent these messages, which is how we have the logs. It also means we have a stake in the answer, so weigh that as you read.

How we ran the study

The sample was 30 Free and Premium X accounts run by our paying customers. Each one sent first messages to people who did not follow it, and we logged them for 21 days, with the totals recorded on 1 September 2026. Sends from one account were spaced 3 to 5 minutes apart.

We counted a refusal each time X answered a send with a limit notice or an error code, and a failed retry counted again. To time the reset, we measured from the first cold DM of a batch to the next send X accepted after a refusal. The retry figures come from our founder's own workspace in the week before 3 September 2026.

All of it ran under the same rule as any other tool. In X's own words, "You may not send unsolicited Direct Messages in a bulk or automated manner". That line covers our software as much as anyone's.

Next send went through about 25 hours later

Across the sample, the next first message X accepted came 24.7 to 27 hours after the batch's first send. Our software tried again about once an hour back then, so the cap may have opened up to an hour sooner. A later check, on 26 September 2026, put X's count at about 24 and a half hours. X's help page only says "in a 24-hour period", and nothing in our logs pointed at midnight.

In practice, a batch that starts at 9 in the morning can start again between about 9:45 and noon the next day. The window rolls from your own first message, so two accounts that start at different hours also reopen at different hours. The X DM limit guide lists the cap for each plan.

At 35 a day, 226 means the cap

Paid accounts that hit their cap got error 226, the message that reads "This request looks like it might be automated". All 23 accounts that stopped at exactly 35 first messages in a day got it on the next try. The 3 free accounts that reached 7 did not get it. Each got a polite notice that it had hit its daily limit.

That matters because X's own error list calls 226 an automation flag, which sounds far worse than a full counter. So count before you worry. If an account shows 226 after exactly 35 first messages, it is almost certainly the cap. Well under 35 is a different case, and what error 226 means on X covers it.

Retrying a refused send almost never works

Of 662 sends that our founder's workspace retried in a single week, to leads X had already refused, 11 got through. The other 651 were refused again, and each one took a sending slot that could have gone to someone new.

That week is the reason we changed our sending. Since 5 September 2026, a lead X turned down once is skipped for good, and the account moves on to the next name on the list.

X refused 1,078 sends but suspended nobody

None of the 30 accounts was suspended in the 21 days. X did ask for a phone or puzzle check 9 times in the study, each in an account's first week, and each cleared in under 24 hours.

Our wider logs cover a longer stretch. From 2 July to 5 October 2026, none of the 46 X accounts Xsender ran was suspended. 9 had a temporary phone or puzzle check. That is a record, not a promise. These accounts sent up to their caps and rotated a few message versions, and the record covers nothing beyond that pattern.

What the study cannot tell you

No account in the sample was on Premium Plus or Premium Business, so we have not seen what happens at 75 or 500 a day. Our sends were also spaced minutes apart, which means we have not tested bursts or any hourly limit, and X publishes none.

The study counted what X allowed, not who answered. For reply rates and the other numbers people quote about X DMs, our page of X cold DM statistics gathers each one with its source.

Our sending now runs on these findings: it times each account's window from its own first send and skips any lead X has refused.

FAQ

Is there an hourly DM limit on X?

X publishes no hourly DM limit. The study spaced sends 3 to 5 minutes apart, so it did not test what fast bursts do.

Does a free X account get error 226 at the DM limit?

Not in the study. All 3 free accounts that hit 7 first messages got a polite daily limit notice, and only paid accounts at their cap got error 226.

Can I use these numbers in my own article?

Yes. Quote each figure with the study name and date, and link to this page or the CSV file so readers can check the totals.

Try it on your own X accounts. See plans and the 7-day trial

Sources and method

Sources and dates
  • Caps by plan and the "24-hour period" wording: X's help page on Chat, archived 4 August 2026, read again 5 October 2026.
  • Error 226 and its wording: X's API error code page, last static copy 17 May 2022.
  • The DM rule: X's automation rules, updated April 2026.
  • Sample, reset window, error 226, refusals, checks and suspensions: our production logs, 30 accounts over 21 days, recorded 1 September 2026.
  • Retries: our founder's own workspace, the week before 3 September 2026. Hourly retries during the study and the 26 September 2026 check of 180 refusals: our sending code and its notes.
  • The later record: our production logs, all workspaces, 2 July to 5 October 2026.