1. Resources
  2. Detection rules
  3. MFA Fatigue

MFA Fatigue detection rules

An attacker who already has a user's password triggers MFA prompts again and again until the user approves one. Denied prompts and user reports of suspicious prompts are the traces in Microsoft Entra ID.

Platform
Microsoft Entra ID
Severity
Medium
Log source
Microsoft Entra ID sign-in logs, Microsoft Entra ID Protection risk detections, Microsoft Entra ID audit logs
Last updated
Rule formats
Sigmaand queries for: SplunkMicrosoft SentinelIBM QRadar
What an attacker needs
The user's password, and an MFA method on the account that can be approved without proof of who started the sign-in, such as a phone call.
Impact
A fully authenticated session as the user that Conditional Access treats as having satisfied MFA, usually followed by registering an attacker-controlled method.
Response
Reset the password of the user at once, even if every prompt was denied, and revoke the sessions of the user if a prompt was approved.
Mitigation
Move users, administrators first, to phishing-resistant methods, which have no prompt that can be approved by mistake.
Risk detection types
authenticatorPhishinguserReportedSuspiciousActivity
Audit activities
Suspicious activity reportedFraud reported - user is blocked for MFAFraud reported - no action taken

How the attack works

MFA fatigue, also called MFA bombing or push spam, starts after the attacker has a user's password. Every sign-in attempt with that password makes Microsoft Entra ID send a push notification or place a phone call to the user's registered device. The attacker repeats the attempt, sometimes for hours or at night, and often adds a message or a call that pretends to come from IT support, until the user approves a prompt to make it stop.

Number matching, which is enabled for all Microsoft Authenticator push notifications, means a blind tap no longer approves a sign-in: the user has to type the number shown on the attacker's screen. That raised the cost of the attack but did not end it. Attackers now pair the prompts with social engineering to obtain the number, or go after methods that only need an approval, such as a phone call.

Each attempt is its own entry in the sign-in log. When the user rejects the prompt, the sign-in fails with error 500121 and its authentication details give the reason, user declined the authentication. If the "Report suspicious activity" setting is enabled, the user can also report the prompt from Microsoft Authenticator or during the phone call. Microsoft Entra ID then writes Suspicious activity reported to the audit log and, with Microsoft Entra ID P2, raises the "User reported suspicious activity" risk detection, which sets the user to high risk. A threshold rule on this page counts error 500121 per user, which also covers prompts that timed out or were answered with a wrong code. Where the risk detection log is collected, two more rules read "User reported suspicious activity" and "Suspicious MFA authentication approval", which Microsoft raises for a prompt that was approved under suspicious circumstances.

What an attacker needs

  • The user's password (phished, sprayed, reused from a breach or bought).
  • An MFA method on the account that can be approved without proof of who started the sign-in, such as a phone call, or a user who can be talked into entering the number of a push notification.

Impact

A fully authenticated session as the user, which Conditional Access treats as having satisfied MFA. The usual next steps are registering an MFA method the attacker controls, reading mail and files, and looking for a way to higher privileges.

Detection rules

MFA Prompt Denied by the User

Detects a sign-in whose authentication details record that the user declined the MFA request (result detail "user declined the authentication"). The user received a prompt and rejected it, which usually means someone else entered the user's password. Microsoft recommends alerting on every denied MFA request. The sign-in itself fails with error 500121.

LevelMediumStatusExperimentalLog sourceMicrosoft Entra ID sign-in logs
title: MFA Prompt Denied by the User
id: 2adff64f-6173-4f4a-961b-243a5287add9
status: experimental
description: >-
  Detects a sign-in whose authentication details record that the user declined
  the MFA request (result detail "user declined the authentication"). The
  user received a prompt and rejected it, which usually means someone else
  entered the user's password. Microsoft recommends alerting on every denied
  MFA request. The sign-in itself fails with error 500121.
references:
  - https://learn.microsoft.com/en-us/entra/identity/authentication/howto-mfa-reporting
  - https://learn.microsoft.com/en-us/entra/architecture/security-operations-user-accounts
  - https://learn.microsoft.com/en-us/azure/azure-monitor/reference/tables/signinlogs
  - https://attack.mitre.org/techniques/T1621/
author: Forestall
date: 2026-11-14
tags:
  - attack.credential-access
  - attack.t1621
logsource:
  product: azure
  service: signinlogs
detection:
  selection:
    AuthenticationDetails|contains: user declined the authentication
  condition: selection
falsepositives:
  - A user who denies a prompt by mistake, or denies a prompt caused by one of their own devices or applications retrying in the background. Confirm with the user; the source IP address of the sign-in settles it.
level: medium

False positives

  • A user who denies a prompt by mistake, or denies a prompt caused by one of their own devices or applications retrying in the background. Confirm with the user; the source IP address of the sign-in settles it.

Repeated MFA Failures for One User

Counts, per user, the sign-ins that failed with error 500121 ("Authentication failed during strong authentication request"). Microsoft's security operations guides list this code among the interrupts to monitor and say that it can be an indication that an attacker has the password for the account but cannot pass the MFA challenge. One such failure is a user who ignored or mistyped a prompt. Five or more within one hour for the same user means that someone keeps starting sign-ins that reach the second factor, which is what an MFA fatigue attack looks like while the user holds out. Unlike the rule for denied prompts, this count also includes prompts that timed out or were answered with a wrong code. The threshold and the window are starting points to tune against your own baseline. The rule reads the interactive sign-in log, so run the same count over the non-interactive user sign-in log (AADNonInteractiveUserSignInLogs in Sentinel) as well.

Fires when one UserPrincipalName has at least 5 matching events within 1 hour.

Threshold ruleLevelMediumStatusExperimentalLog sourceMicrosoft Entra ID sign-in logs
title: Sign-In Failed at the MFA Step (Repeated MFA Failure Base Event)
id: e3a3338c-cc27-4f34-8650-2ca3ace64fa1
name: entra_sign_in_failed_at_mfa_step
status: experimental
description: >-
  Base event of the threshold rule in this file. A Microsoft Entra sign-in
  that failed with error 500121, "Authentication failed during strong
  authentication request": the sign-in reached the MFA challenge and the
  challenge was not completed.
logsource:
  product: azure
  service: signinlogs
detection:
  selection:
    ResultType: '500121'  # string column
  condition: selection
level: low
---
title: Repeated MFA Failures for One User
id: 2b8a2e07-3ad1-4918-9456-f62855d2e48d
status: experimental
description: >-
  Counts, per user, the sign-ins that failed with error 500121
  ("Authentication failed during strong authentication request"). Microsoft's
  security operations guides list this code among the interrupts to monitor
  and say that it can be an indication that an attacker has the password for
  the account but cannot pass the MFA challenge. One such failure is a user
  who ignored or mistyped a prompt. Five or more within one hour for the same
  user means that someone keeps starting sign-ins that reach the second
  factor, which is what an MFA fatigue attack looks like while the user holds
  out. Unlike the rule for denied prompts, this count also includes prompts
  that timed out or were answered with a wrong code. The threshold and the
  window are starting points to tune against your own baseline. The rule
  reads the interactive sign-in log, so run the same count over the
  non-interactive user sign-in log (AADNonInteractiveUserSignInLogs in
  Sentinel) as well.
references:
  - https://learn.microsoft.com/en-us/entra/architecture/security-operations-user-accounts
  - https://learn.microsoft.com/en-us/entra/architecture/security-operations-privileged-accounts
  - https://attack.mitre.org/techniques/T1621/
author: Forestall
date: 2026-11-18
tags:
  - attack.credential-access
  - attack.t1621
correlation:
  type: event_count
  rules:
    - entra_sign_in_failed_at_mfa_step
  group-by:
    - UserPrincipalName
  timespan: 1h
  condition:
    gte: 5
falsepositives:
  - A user with a new phone, a broken authenticator app or no network on the phone, who keeps retrying. One call to the user settles it.
  - Applications and devices that retry a sign-in in the background and reach an MFA prompt nobody sees.
  - The values of five failures and one hour are starting points, not tested limits. Use a lower threshold for privileged accounts.
level: medium

False positives

  • A user with a new phone, a broken authenticator app or no network on the phone, who keeps retrying. One call to the user settles it.
  • Applications and devices that retry a sign-in in the background and reach an MFA prompt nobody sees.
  • The values of five failures and one hour are starting points, not tested limits. Use a lower threshold for privileged accounts.

ID Protection Detected a Suspicious MFA Approval

Detects the Microsoft Entra ID Protection risk detection "Suspicious MFA authentication approval" (riskEventType authenticatorPhishing). Microsoft raises it for sign-ins that used a password and MFA with the Microsoft Authenticator app when unfamiliar properties such as the ASN, browser, device and GPS location point to social engineering or phishing. It compares the location of the device that asked for the sign-in with the location of the device that approved it, to tell a legitimate sign-in from a session an attacker started, and it flags the sign-in as high risk. This is the case the other rules on this page cannot see: the user approved the prompt. The detection is calculated in real time and needs Microsoft Entra ID P2. Risk detections reach a SIEM through the UserRiskEvents diagnostic setting of Microsoft Entra ID.

LevelHighStatusExperimentalLog sourceMicrosoft Entra ID Protection risk detections
title: ID Protection Detected a Suspicious MFA Approval
id: 541f15a1-65c5-4e12-8e8a-b0f9178bd995
status: experimental
description: >-
  Detects the Microsoft Entra ID Protection risk detection "Suspicious MFA
  authentication approval" (riskEventType authenticatorPhishing). Microsoft
  raises it for sign-ins that used a password and MFA with the Microsoft
  Authenticator app when unfamiliar properties such as the ASN, browser,
  device and GPS location point to social engineering or phishing. It compares
  the location of the device that asked for the sign-in with the location of
  the device that approved it, to tell a legitimate sign-in from a session an
  attacker started, and it flags the sign-in as high risk. This is the case
  the other rules on this page cannot see: the user approved the prompt. The
  detection is calculated in real time and needs Microsoft Entra ID P2. Risk
  detections reach a SIEM through the UserRiskEvents diagnostic setting of
  Microsoft Entra ID.
references:
  - https://learn.microsoft.com/en-us/entra/id-protection/concept-identity-protection-risks
  - https://learn.microsoft.com/en-us/entra/id-protection/howto-identity-protection-investigate-risk
  - https://learn.microsoft.com/en-us/azure/azure-monitor/reference/tables/aaduserriskevents
  - https://attack.mitre.org/techniques/T1621/
author: Forestall
date: 2026-11-18
tags:
  - attack.credential-access
  - attack.t1621
logsource:
  product: azure
  service: riskdetection
detection:
  selection:
    RiskEventType: authenticatorPhishing
  condition: selection
falsepositives:
  - None documented by Microsoft. A user who knowingly approves a sign-in that was started on a device in another place would look the same, so confirm with the user which sign-in was approved.
level: high

False positives

  • None documented by Microsoft. A user who knowingly approves a sign-in that was started on a device in another place would look the same, so confirm with the user which sign-in was approved.

User Reported a Suspicious MFA Prompt

Detects a user reporting an MFA prompt they did not start. With the "Report suspicious activity" setting enabled, Microsoft Entra ID writes the activity "Suspicious activity reported" to the audit log; the two "Fraud reported" activities come from the older fraud alert feature that it replaced. A report means the user's password is known to someone else.

LevelHighStatusExperimentalLog sourceMicrosoft Entra ID audit logs
title: User Reported a Suspicious MFA Prompt
id: 462edd6e-6aeb-411d-bc66-490c67a0f5b2
status: experimental
description: >-
  Detects a user reporting an MFA prompt they did not start. With the "Report
  suspicious activity" setting enabled, Microsoft Entra ID writes the activity
  "Suspicious activity reported" to the audit log; the two "Fraud reported"
  activities come from the older fraud alert feature that it replaced. A
  report means the user's password is known to someone else.
references:
  - https://learn.microsoft.com/en-us/entra/identity/authentication/howto-mfa-mfasettings
  - https://learn.microsoft.com/en-us/entra/identity/monitoring-health/reference-audit-activities
  - https://attack.mitre.org/techniques/T1621/
author: Forestall
date: 2026-11-14
tags:
  - attack.credential-access
  - attack.t1621
logsource:
  product: azure
  service: auditlogs
detection:
  selection:
    OperationName:
      - Suspicious activity reported
      - Fraud reported - user is blocked for MFA
      - Fraud reported - no action taken
  condition: selection
falsepositives:
  - A user who chooses the report option by mistake. Confirm with the user before dismissing the risk.
level: high

False positives

  • A user who chooses the report option by mistake. Confirm with the user before dismissing the risk.

ID Protection Recorded a User Report of a Suspicious MFA Prompt

Detects the Microsoft Entra ID Protection risk detection "User reported suspicious activity" (riskEventType userReportedSuspiciousActivity). Microsoft raises it when a user denies an MFA prompt and reports it as suspicious, and notes that a prompt the user did not start can mean that the credentials are compromised. It is the risk log counterpart of the audit rule on this page and is useful where the risk detections are collected and the audit activity is not. The detection needs the "Report suspicious activity" setting to be turned on and Microsoft Entra ID P2, and it is calculated offline, so it arrives after the report. Risk detections reach a SIEM through the UserRiskEvents diagnostic setting of Microsoft Entra ID.

LevelHighStatusExperimentalLog sourceMicrosoft Entra ID Protection risk detections
title: ID Protection Recorded a User Report of a Suspicious MFA Prompt
id: 17744021-d5aa-4f06-b190-6ffb55ca4730
status: experimental
description: >-
  Detects the Microsoft Entra ID Protection risk detection "User reported
  suspicious activity" (riskEventType userReportedSuspiciousActivity).
  Microsoft raises it when a user denies an MFA prompt and reports it as
  suspicious, and notes that a prompt the user did not start can mean that the
  credentials are compromised. It is the risk log counterpart of the audit
  rule on this page and is useful where the risk detections are collected and
  the audit activity is not. The detection needs the "Report suspicious
  activity" setting to be turned on and Microsoft Entra ID P2, and it is
  calculated offline, so it arrives after the report. Risk detections reach a
  SIEM through the UserRiskEvents diagnostic setting of Microsoft Entra ID.
references:
  - https://learn.microsoft.com/en-us/entra/id-protection/concept-identity-protection-risks
  - https://learn.microsoft.com/en-us/entra/identity/authentication/howto-mfa-mfasettings
  - https://learn.microsoft.com/en-us/azure/azure-monitor/reference/tables/aaduserriskevents
  - https://attack.mitre.org/techniques/T1621/
author: Forestall
date: 2026-11-18
tags:
  - attack.credential-access
  - attack.t1621
logsource:
  product: azure
  service: riskdetection
detection:
  selection:
    RiskEventType: userReportedSuspiciousActivity
  condition: selection
falsepositives:
  - A user who reports a prompt that their own device or application caused. Check the sign-in behind the prompt before closing the report.
level: high

False positives

  • A user who reports a prompt that their own device or application caused. Check the sign-in behind the prompt before closing the report.

Detection logic

Patterns that need more than one event, a baseline or knowledge of your own environment. Each is a hypothesis to build and tune in your SIEM, not a ready rule.

Failed MFA prompts followed by a successful sign-in for the same user

Detection hypothesisSequence of events

MFA fatigue ends when the user gives in. A run of sign-ins that fail at the MFA step, followed by one that succeeds from the address that caused the failures, is the moment the attack worked. The failures alone are an attempt, and the success alone is an ordinary sign-in.

  1. Sign-in log: three or more sign-ins with ResultType 500121 for one UserPrincipalName within 60 minutes. Keep the IPAddress values.
  2. Within 30 minutes of the last failure: a sign-in with ResultType 0 for the same UserPrincipalName.
  3. Alert when the IPAddress of the successful sign-in is one of the addresses of the failures, and raise the priority when the user denied or reported a prompt in that period.

Events500121

LimitsA user who struggles with a new phone and then succeeds from the same place looks the same, so check whether the address and the device are the user's own. The counts and time spans are starting points. An approval of the very first prompt has no failures before it and does not match.

Triage

  • Ask the user whether they started the sign-in. A prompt nobody asked for means the password is known to someone else.
  • Compare the IP address, location and client of the denied sign-in with the user's normal sign-ins, and count the prompts the same source caused (error 500121), for this user and for others.
  • Look for a successful sign-in from the same source after the denials. An approval that follows a series of denials means the attack worked.
  • If the password is exposed, reset it, revoke the user's sessions and review the MFA methods registered on the account since the first prompt.

Response

Once the attack is confirmed:

  1. Reset the password of the user at once, even if every prompt was denied. The prompts stop only when the password the attacker holds no longer works.
  2. If a prompt was approved, revoke the sessions of the user and block sign-in until the account is checked.
  3. Remove every authentication method registered on the account since the first unexpected prompt, and check the remaining ones with the user.
  4. With Microsoft Entra ID P2, make sure the user risk raised by the report is closed by a secure password change and not simply dismissed.
  5. Look for the same source IP address prompting other users, and handle each of them in the same way.

Mitigation

  • Move users, administrators first, to phishing-resistant methods (FIDO2 security keys, passkeys, Windows Hello for Business, certificate-based authentication). They have no prompt that can be approved by mistake.
  • Enable "Report suspicious activity" in the authentication methods policy, so that a reported prompt marks the user as high risk, and with Microsoft Entra ID P2 use a user risk policy that blocks access or requires a secure password change at high risk.
  • Show the application name and the geographic location in Microsoft Authenticator notifications, so users can see that a prompt is not theirs.
  • Remove phone call and SMS methods where they are not needed.
  • Teach users that an unexpected prompt means someone knows their password. The right reaction is to deny, report and change the password.

Frequently asked questions

Did number matching end MFA fatigue attacks?

No. A blind tap no longer approves a sign-in, because the user has to type the number shown on the attacker's screen. Attackers now pair the prompts with social engineering to obtain the number, or go after methods that only need an approval, such as a phone call.

The user denied the prompt and the sign-in failed, so why does it still matter?

A prompt is sent only after the attacker entered the correct password. A prompt the user did not start therefore means the password is known to someone else, even though this attempt failed. Reset the password, or the prompts continue until one is approved.

What has to be enabled for the report rule to produce events?

The "Report suspicious activity" setting in the authentication methods policy. With it, a user can report a prompt from Microsoft Authenticator or during the phone call, and Microsoft Entra ID writes `Suspicious activity reported` to the audit log. With Microsoft Entra ID P2 the report also sets the user to high risk.

Can these rules see an attack in which the user approves the first prompt?

Only one of them can, and it needs Microsoft Entra ID P2. The sign-in and audit rules read failed and denied prompts and user reports, and an approved prompt is a successful sign-in that matches none of them. The rule for the risk detection "Suspicious MFA authentication approval" is the exception, because Microsoft raises that detection for sign-ins whose properties point to an approval obtained by social engineering or phishing. Where that detection is not available, an approval without any earlier failure or report leaves nothing these rules can read.

References

We respect your privacy

We use cookies to keep this site secure and working properly. With your permission, we also use optional cookies to understand usage and improve the experience. Cookie Policy

You can change your choice at any time.