Skip to content

Est. 2011ยทMicrosoft Partner ID 7033487ยทDelivery under 3 minยทSupport 7 days a week

Your vault is empty.

Free Fix Event ID 1035

Event ID 1035: inbound authentication failed on an Exchange receive connector

10 min read Updated October 5, 2026 Exchange Server

Fix it now

An inbound SMTP session offered credentials and was refused. Microsoft publishes no description for this event, so read the entry itself: the connector that answered, the mechanism attempted and the source address are the three fields that explain almost every case.

Run these in the Exchange Management Shell on the server that logged the event

Get-ReceiveConnector | Format-List Name,TransportRole,Bindings,RemoteIPRanges,AuthMechanism,PermissionGroups
Set-ReceiveConnector -Identity '<connector>' -ProtocolLoggingLevel Verbose
  1. Open the Application log, filter the source to MSExchangeTransport, and write down the connector name, the mechanism and the source IP address from the entry.
  2. Work out which connector covers that source address most narrowly. Microsoft documents that where remote IP ranges overlap, the connector with the most specific match to the connecting IP address is used.
  3. If the client landed on the internet-facing connector on port 25, move it. Devices that submit with a username and password belong on the client submission connector on 587.
  4. If the device cannot negotiate STARTTLS, give it its own connector scoped to its address alone, in the Front End Transport service, with only the settings that device can use.
  5. Reproduce the failure with verbose protocol logging on and read the session in the SmtpReceive log under the transport logging path.

Turn verbose protocol logging back off when you have finished. Left on, it is one of the most reliable ways to fill a volume and cause a back pressure incident weeks later.

If the device submits and no new entry appears, you are done. If not, the next section explains how Exchange chooses a connector and what each property decides.

Why it happens

Every receive connector has two scopes. Bindings is the local address and port it listens on. RemoteIPRanges is the set of source addresses it will serve. Microsoft documents that multiple connectors on the same server may have overlapping remote ranges as long as one range is completely overlapped by another, and that when they overlap, the connector with the most specific match to the connecting server’s IP address is used. From that moment the session is governed by that connector alone, which is why a device you configured perfectly can still fail: something else out-scoped it.

Two properties then decide whether authentication is even possible. AuthMechanism controls what the connector advertises and accepts, with documented values None, Tls, Integrated, BasicAuth, BasicAuthRequireTLS, ExchangeServer and ExternalAuthoritative. PermissionGroups controls what the authenticated party may then do, with values None, AnonymousUsers, ExchangeUsers, ExchangeServers, ExchangeLegacyServers, Partners and Custom. A client offering basic credentials to a connector that advertises only anonymous submission is refused before its password is ever checked.

TransportRole decides where the connector lives: FrontendTransport is where client and external SMTP connections arrive, HubTransport is where Exchange server and proxied client connections occur. Putting a client submission connector in the wrong service is a common way to build something that never receives the session it was created for.

A different connector out-scoped the one you configured

You have this one if The connector named in the event is not the connector you built for this device.

  1. List every connector with its bindings and remote ranges, then find each one whose range contains the source address.
  2. Give the device its own connector scoped to its address alone: New-ReceiveConnector -Name AppSubmission -TransportRole FrontendTransport -Bindings 0.0.0.0:587 -RemoteIPRanges 10.0.5.20 -Usage Client.
  3. Confirm the new connector is now the most specific match with Get-ReceiveConnector | Format-List Name,RemoteIPRanges.

Microsoft documents -Usage Client as requiring RemoteIPRanges, and as applying permission group ExchangeUsers with Tls, BasicAuth, BasicAuthRequireTLS and Integrated. That is the standard client submission shape.

The connector will not accept the mechanism the client offers

You have this one if The mechanism named in the event is one the connector does not advertise, or it is basic on a connector that requires encryption first.

  1. Read the current settings: Get-ReceiveConnector -Identity AppSubmission | Format-List AuthMechanism,PermissionGroups,Fqdn.
  2. Fix the device rather than the server where you can. Submitting on 587 with STARTTLS and basic credentials is what almost every appliance supports.
  3. If the device genuinely cannot do TLS, put it on its own internally bound, address-scoped connector and accept basic authentication there only.

Never remove the requirement for TLS from a connector reachable from the internet. Basic credentials on an unencrypted internet-facing listener are harvested within days.

The credentials themselves are being rejected

You have this one if The same account fails from every source, and a second account works from the same device.

  1. Check the account state with Get-ADUser <name> -Properties LockedOut,PasswordExpired,Enabled.
  2. Try both credential formats. Some devices send only the down-level domain and user form, others only the user principal name.
  3. Confirm the account has a mailbox. An account with no mailbox, or a disabled one, cannot submit as itself.
  4. If the password was rotated, update every device that used it. A single stale copy produces a steady stream of these entries and can lock the account repeatedly.

TLS is not being established, so the session is refused

You have this one if The problem started when a certificate was replaced, renewed or allowed to expire.

  1. Check the certificate assigned to SMTP: Get-ExchangeCertificate | Format-List Thumbprint,Services,Subject,NotAfter.
  2. If the services list does not include SMTP, assign it: Enable-ExchangeCertificate -Thumbprint <thumbprint> -Services SMTP.
  3. Confirm the name the connector announces matches a name on the certificate, and set it with Set-ReceiveConnector -Identity <name> -Fqdn mail.example.com if not.

Full reference

The connector properties that decide the session

Property Documented values What it decides
Bindings IPv4 or IPv6 address and TCP port; 0.0.0.0 or [::] means all local addresses Which sessions the connector can hear
RemoteIPRanges Single address, range, or CIDR block Which sources it will serve, and how specific its claim is
AuthMechanism None, Tls, Integrated, BasicAuth, BasicAuthRequireTLS, ExchangeServer, ExternalAuthoritative What it advertises and accepts
PermissionGroups None, AnonymousUsers, ExchangeUsers, ExchangeServers, ExchangeLegacyServers, Partners, Custom What the authenticated party may do
TransportRole FrontendTransport, HubTransport Which transport service owns it
ProtocolLoggingLevel None, Verbose Whether the session is written to the protocol log

What the usage types set for you

Rather than assembling permission groups and authentication mechanisms by hand, New-ReceiveConnector -Usage applies a documented combination. Client gives permission group ExchangeUsers with Tls, BasicAuth, BasicAuthRequireTLS and Integrated. Internet gives AnonymousUsers with Tls. Internal gives ExchangeLegacyServers and ExchangeServers with Tls and ExchangeServer. Partner gives Partners with Tls. Custom gives no permission group and Tls only. Client, Internal, Partner and Custom all require RemoteIPRanges; Internet, Partner and Custom all require Bindings.

The defaults that setup creates are load bearing. Rather than loosening one of them to accommodate one appliance, scope a new connector to that appliance’s address and leave the defaults alone.

The codes that appear alongside

421 4.7.0 and 451 4.7.0 are not published by Microsoft. RFC 3463 defines the subject digit 7 as security or policy, and X.7.0 as a security-related problem that no more specific detail code describes; the leading 4 in both makes them temporary. The difference between them is the SMTP reply code rather than the detail: 421 closes the transmission channel, 451 reports a temporary local failure without ending the session. Neither tells you why, which is the point of reading the event and the protocol log instead.

Where the logs are

  • Protocol logs for receive connectors sit under the transport logging tree. Microsoft’s antivirus exclusion list documents %ExchangeInstallPath%TransportRoles\Logs as holding mail flow and transport pipeline logs in subfolders, and %ExchangeInstallPath%Logging as holding many other Exchange log types.
  • Turn logging on per connector with Set-ReceiveConnector -ProtocolLoggingLevel Verbose, and off again with None. Microsoft documents None as the default.
  • Correlate the protocol log session with the event by timestamp and source address, not by connector name alone.
  • If a second event ID appears alongside, match it the same way. Microsoft publishes no description for these transport authentication events, so the entry text on your server is the authority.

When a network change breaks every device at once

The source address Exchange sees is whatever arrives at the server. If a load balancer or firewall starts translating client addresses, every connector match changes at the same moment, and devices that used to reach their own connector land on whichever one covers the new address. That is why this failure often appears the morning after a network change nobody thought was related to mail. Check what address the event names before you touch any connector.

Every code this article covers

Code What it points at Source
Event ID 1035 No description is published by Microsoft. Read the entry itself: it identifies the receive connector that answered, the authentication mechanism attempted and the source IP address of the refused session not published by the vendor
Event ID 1009 No description is published by Microsoft. Correlate it with the 1035 entry by timestamp and source address and read what your own entry says not published by the vendor
421 4.7.0 A temporary security or policy failure returned as the server closes the transmission channel. RFC 3463 defines X.7.0 as a security problem no more specific code describes RFC 3463
451 4.7.0 The same temporary security or policy class returned without ending the session outright RFC 3463

Confirm the fix worked

  1. Reproduce the submission and confirm no new entry appears for that source address.
  2. Read the verbose protocol log for the session and confirm the authentication command was accepted and the message queued.
  3. Confirm with Get-ReceiveConnector | Format-List Name,RemoteIPRanges that the connector you intended is the most specific match for the client.
  4. Send a test message end to end and find it with Get-MessageTrackingLog -Sender <address>, then set ProtocolLoggingLevel back to None.

Questions people ask about this

Does fixing this cost anything?

No. Connector scoping, authentication mechanisms and permission groups are all configuration, and every tool you need ships with Exchange. The only licensing question nearby is that an application authenticating as a user needs a real mailbox to authenticate as.

Where is the official description of this event?

There is not one. Microsoft publishes no event description for it, which is why this article works from the connector properties and the protocol log rather than from the number.

Should I just allow anonymous relay?

Only scoped to specific source addresses, on a connector not reachable from the internet. An open relay will be found and used, and your sending domains will end up on blocklists.

Why did this start after a network change?

Because the source address Exchange sees is the one that arrives at the server. If a load balancer or firewall started translating client addresses, every connector match changed at once.

Can I just stop the events being logged?

You can lower the logging level, but the submissions still fail. The event is the symptom, not the problem, and silencing it leaves the next person with nothing to work from.

Related error codes

Was this article helpful?

Your feedback helps us improve our documentation.

Related articles

License Error Event ID 9518 with error -1808: the log volume filled and the store dismounted Free Fix Event ID 2059: log shipping stops when the DAG replication network breaks License Error Event ID 9646: a client blew past the Exchange session or object limit Free Fix 0x80070643 during an Exchange CU: setup fails and rolls the server back
โ† Back to Knowledge Base