Fix it now
The move read the source mailbox, found items it could not copy, and stopped because the count passed the tolerance you set. The default tolerance is zero, so a single bad item is enough. Raising the limit is a legitimate fix, but it is a decision to abandon those items, so capture what they are before you change anything.
$stats = Get-MailboxStatistics user@contoso.com -IncludeMoveReport
$report = $stats.MoveHistory[0].Report
$report.BadItems
Get-MoveRequest user@contoso.com | Set-MoveRequest -BadItemLimit 10
Resume-MoveRequest user@contoso.com
- For a batch-based migration read the skipped items instead:
Get-MigrationUserStatistics -Identity user@contoso.com -IncludeSkippedItems | Select-Object -ExpandProperty SkippedItems | Format-List DateReceived, Subject. - Save that output somewhere durable. Once the move completes the detail becomes much harder to recover, and it is the only record of what the owner lost.
- Set the limit slightly above the confirmed count rather than at a round number. Microsoft recommends 10 or lower.
- If the count is high enough that a sensible limit will not cover it, repair the source instead:
New-MailboxRepairRequest -Mailbox user@contoso.com -CorruptionType <type>, then retry the move.
BadItemLimit and LargeItemLimit do not exist on Set-MoveRequest in Exchange Online. If you are working in the cloud service, the tolerance is not yours to set on the request and the repair route is the one you have.
If the request moves past the point it stopped at and the skipped list is saved, you are done. If the counts climb every time you resume, the next section explains what is actually failing.
Why it happens
The replication service reads each item, converts it into the form the target store expects, and writes it. An item is bad when that pipeline cannot complete: the item is structurally damaged in the source database, a property is malformed, an attachment cannot be read, or a security descriptor references something that no longer resolves. These are faults in the data, which is why retrying does not help and why the service refuses to guess at a repair.
There are two counters, not three. Bad items and large items each have their own limit, and Microsoft is explicit that missing items are counted inside the bad item limit rather than having one of their own. A missing item was listed in the source but was gone when the service read it, which happens routinely when someone deletes mail mid-move. So a mailbox with a few corrupt items and a user tidying their inbox trips the same limit twice as fast as you expected.
Large items are the separate case. Nothing is wrong with the item; it exceeds the maximum message size the target mailbox accepts, and the organisation-wide value applies if the mailbox has none of its own. That distinction matters because the switch Exchange demands at high tolerances is tied to the large item limit on current versions, not to the bad item limit.
The mailbox holds genuinely corrupt items
You have this one if The bad item list names specific items with conversion or property failures, and the same items come back on every attempt.
- Export the bad item list and keep it with your migration records.
- Set a limit slightly above the confirmed count. Microsoft’s recommendation is 10 or lower, and the default is 0.
- Resume and confirm the request progresses past the stopping point.
- After the move, ask the owner to check the affected folders while backups of the source still exist.
Setting the limit to unlimited removes the safety net entirely. If a mailbox needs that, the mailbox or its database deserves attention before the move does.
The items are damaged in a way a repair can fix
You have this one if The counts are high, they are concentrated in one or two folders, and the same folders appear on every attempt.
- Run
New-MailboxRepairRequest -Mailbox <user> -CorruptionType <type>against the source mailbox, choosing the corruption type that matches what the report describes. - Expect access to that mailbox to be disrupted while it runs, and know you cannot stop a repair short of dismounting the database.
- Retry the move once it completes and compare the new bad item count against the old one.
This is the step Microsoft names when too many bad items are detected. It is a better answer than a large tolerance because it recovers items rather than abandoning them.
The items are too large for the target rather than damaged
You have this one if TooManyLargeItemsPermanentException, and the skipped entries are all sizeable messages with big attachments.
- Compare the maximum message size the target accepts with the sizes in the report. The organisation-wide value applies if the mailbox has none of its own.
- Where you can, have owners archive the oversized items before the move rather than losing them.
- If they must be abandoned, raise
LargeItemLimitrather thanBadItemLimit, and add-AcceptLargeDataLossabove 50. On Exchange 2013 and later that switch is required when LargeItemLimit is 51 or higher.
Users are deleting mail while their mailbox is being moved
You have this one if The counts differ between attempts, the affected mailboxes belong to active users, and nothing in the list looks corrupt.
- Check whether the mailbox was in heavy use during the sync window, and schedule the final synchronisation for a quieter period.
- Tolerate a small missing item count. These are items the user removed themselves, and they still count against the bad item limit.
- If the counts are large and the user denies deleting anything, look for a retention policy or a client add-in doing it for them.
The store is refusing operations rather than individual items failing
You have this one if StoragePermanentException appears, or the failure arrives immediately rather than after a period of progress.
- Confirm the source database is mounted:
Get-MailboxDatabase -Status | Format-Table Name,Mounted, and check free space on the database and log volumes. - Confirm the mailbox is not disconnected or mid-move:
Get-MailboxStatistics -Identity <user> | Format-List Database,DisconnectReason. - Resolve the store fault before touching item limits. A limit cannot fix a database problem.
Full reference
Which limit applies where
| Parameter | Default | What it counts | Where it exists |
|---|---|---|---|
BadItemLimit |
0 | Corrupt items, plus missing items | On-premises Exchange. Not available on Set-MoveRequest in Exchange Online |
LargeItemLimit |
0 | Items over the target’s maximum message size | On-premises Exchange. Not available on Set-MoveRequest in Exchange Online |
AcceptLargeDataLoss |
off | Required above 50 on the large item limit for Exchange 2013 and later | On-premises Exchange |
Set-MigrationBatch -BadItemLimit |
0 | The same tolerance applied to a batch | On-premises only, and deprecated from the cloud-based service |
On Exchange 2010 the AcceptLargeDataLoss switch was tied to BadItemLimit instead. Anything written for that era, including a lot of migration runbooks still in circulation, gets the coupling wrong for the version you are running.
Choosing a number you can defend
| What the list looks like | Reasonable response |
|---|---|
| A handful of calendar or contact items | Raise the limit modestly, complete the move, tell the owner what went |
| Hundreds of items from one folder | Repair the mailbox before moving it; the damage is local and may be recoverable |
| Everything is oversized rather than corrupt | This is a large item question, not a bad item one. Raise the other limit |
| Items vanish between attempts | Missing items from ordinary user deletion, counted against the bad item limit |
| The counts climb every time you resume | Stop. Something is actively damaging the mailbox or the database |
Getting the evidence out
| Command | What it gives you |
|---|---|
$stats = Get-MailboxStatistics <user> -IncludeMoveReport |
Loads the move history, including completed moves |
$stats.MoveHistory[0].Report.BadItems |
The bad item list itself, which is what you are meant to read |
Get-MigrationUserStatistics <user> -IncludeSkippedItems |
Skipped items for a batch-based migration |
Get-MigrationUserStatistics <user> -LimitSkippedItemsTo <n> |
Caps how many skipped items are returned, for very long lists |
Get-MailboxStatistics <user> | Format-List TotalItemSize,ItemCount |
Counts for a before and after comparison |
What a tolerance actually costs
Raising the limit tells the replication service to discard items rather than fail. They are not copied to the target and no second attempt is made. That is a defensible decision for five damaged calendar entries and an indefensible one for four hundred messages nobody has looked at. The difference between the two is entirely whether you read the list, and the list is only easy to get before the move finishes.
Write down the number you set, the count it was based on, and the date. Six months later, when someone asks why a message from 2019 is not in their mailbox, that record is the difference between an answer and a shrug.
Every code this article covers
| Code | What it points at | Source |
|---|---|---|
TooManyBadItemsPermanentException |
The number of items that could not be copied exceeded the configured bad item limit, and the migration cannot continue until the limit is raised | Microsoft Learn |
TooManyLargeItemsPermanentException |
Too many items exceeded the maximum message size the target mailbox accepts | not published by the vendor |
TooManyMissingItemsPermanentException |
Items listed in the source were gone when the service read them. Microsoft states these are counted in the bad item limit, though they do not increment it | Microsoft Learn |
StoragePermanentException |
The store refused an operation in a way that will not resolve on retry, pointing at database or mailbox state rather than at individual items | not published by the vendor |
Confirm the fix worked
- The bad item or skipped item list is saved outside the migration tooling, with a date on it.
Get-MoveRequestStatisticsshows the status past the point it previously failed.- Source and target item counts from
Get-MailboxStatisticsare consistent, allowing for what you knowingly skipped. - The mailbox owner has opened the folders named in the list and confirmed nothing critical is gone.
- The limit you set is recorded with the count it was based on, not just left on the request.
Questions people ask about this
What limit should I set?
A number slightly above the confirmed count for that mailbox, decided per mailbox. Microsoft recommends 10 or lower. Blanket high limits across a whole batch mean nobody ever finds out what was lost.
Why does Exchange demand an extra switch above a certain number?
Because past that point you are accepting meaningful data loss rather than tolerating a few damaged items. On Exchange 2013 and later the switch is required when the large item limit is 51 or higher; on Exchange 2010 it applied to the bad item limit instead.
Can I recover the skipped items afterwards?
Not from the move. If they still exist in the source mailbox and the source has not been decommissioned, they can sometimes be exported separately. Once the source is gone, only backups help, and only while they last.
Why can I not set BadItemLimit in Exchange Online?
Those parameters are not available on Set-MoveRequest in the cloud service, and the batch equivalents are deprecated there. Repairing the source mailbox is the route that remains, which is arguably the one you wanted anyway.
Is running a mailbox repair safe during business hours?
Access to the mailbox being repaired is disrupted while it runs, and you cannot stop a repair once it has started without dismounting the database. Schedule it for one mailbox at a time rather than starting a database-wide repair on a whim.
