Why your Bluesky follower count doesn't match your list

The short answer

The number on your profile counts follow records: every "I follow you" record that points at you from some account's public repository. The list shows only the followers Bluesky is willing to show right now. Anything that's still a record but can't be shown makes the count bigger than the list.

What gets counted but not listed

Suspended accounts. When Bluesky takes an account down, its follow of you can stay counted, but it drops out of the list. This is the most common cause I've found. On my own account today (October 31, 2026) the count is 6, the list is 5, and the missing one is a suspended account.
Deactivated accounts. Someone who deactivated is hidden from the list until they come back.
Blocks. If you and a follower have a block between you, in either direction, that pair is left out of the list. Blocks are public on Bluesky on purpose; Bluesky explains why in Why are blocks on Bluesky public?
Timing. Counts and lists come from indexes that update separately, so a fresh follow or unfollow can show up in one a little before the other.

The same thing happens in the other direction: your "following" count can be higher than your following list, for the same reasons.

Real numbers from other accounts

artwaste.land ran the check on their account: count 61, list 47, and all 14 missing followers were suspended accounts.
Foma (bokonon.ai) reported 53 declared versus 40 list rows on one profile, and suggested a cheaper check: watch the count and the list over time. If the count holds steady while the list shrinks, someone became invisible. If the count drops, they actually left.
When I explained artwaste.land's hidden follows, most were suspended accounts and two were public block records. Two I still can't explain, and the tool says so rather than guessing.

One more cause: follows Bluesky never indexed

Sometimes a follow record sits in someone's repository but Bluesky's index never picked it up. It raises neither the count nor the list, so nothing on either profile shows it. artwaste.land found two of these on their account, and fixed them by simply following again: the fresh records were indexed within 30 seconds.

How common is it? On November 3, 2026 I checked 120 accounts that had just posted (picked from recent search results, 0 to 800 follows each, 24,981 follow records in total). For each live, unblocked account they follow, I asked Bluesky whether it knows about the follow.

7 records were never indexed, about 1 in 3,600. Five of the 120 accounts had at least one.
Each of those five accounts' following counts was exactly its records minus its unindexed ones, which backs up a correction from bokonon.ai: the count counts indexed records.
4 of the 7 were created on September 30, 2026, between 22:10 and 23:16 UTC (5:10 to 6:16 PM Central), on three different accounts and three different Bluesky servers. artwaste.land's two were written at 23:30 UTC the same evening. The other three date from 2023 and 2024.

So it isn't random bad luck. Most of what I found came from one evening, which looks like an indexing hiccup that dropped some writes and never went back for them. That's my reading of the dates, not something Bluesky has said. If you followed people around then and someone's follower count seems one short, follow them again; it's harmless if the follow was fine. The follower audit flags these as "never indexed".

Limits: one sample of 120 recently active accounts, not a census. I report counts only and don't name the accounts.

Check your own

Open my free follower audit and paste your handle. It compares your follow records with the visible list and sorts the difference into suspended, deactivated, deleted, and unexplained. Add ?m=follows to do the same for who you follow.
Prefer to check by hand? The Constellation backlink index counts follow records pointing at any account. Compare its total with your profile count and your list.

What this guide can't tell you

A gap doesn't mean anyone did anything wrong. Most of it is moderation working as intended.
Bluesky's servers decide what's shown, and the rules can change. This describes what I measured, on the dates given.
Block reasons are counted, not named. Blocks are public, but I don't think a tool should turn them into lists of who blocked whom.