Quick answer
Essential Eight Maturity Level Three (ML3) is the highest level of ASD's Essential Eight Maturity Model. It is designed to resist adaptive, well-resourced adversaries who target a specific organisation and work to evade its controls. ML3 adds the strictest patching timeframes, application control across all servers, phishing-resistant MFA at broad scope, just-in-time administration and active analysis of security logs.
On this page
What ML3 Adds
ML3 includes every ML1 and ML2 requirement and is designed for an adversary who will adapt to your defences. In summary, the main additions are below. Confirm the exact requirements against ASD's current published model.
- Application control: applies to all workstations and servers, covers drivers as well as programs, scripts and libraries, uses Microsoft's recommended application and driver block rules, and the rule set is validated at least annually.
- Patch applications and operating systems: the shortest timeframes apply, including rapid patching where a working exploit exists across a wider range of systems, and applications and operating systems that are no longer supported are removed.
- Microsoft Office macros: only macros from trusted locations, or digitally signed by trusted publishers, can run, and the list of trusted publishers is reviewed.
- User application hardening: hardening is applied more completely across browsers, office productivity suites, PDF software and scripting environments, following ASD and vendor guidance and closing off the legacy components attackers abuse.
- Restrict administrative privileges: privileged access is granted just in time, limited to what each task needs, and credential protection features such as Credential Guard are enabled.
- Multi-factor authentication: phishing-resistant MFA is required for all in-scope users and services, including important data repositories.
- Regular backups: only designated break-glass accounts can modify or delete backups, so even a compromised administrator cannot destroy recovery points.
- Logging: security events are collected centrally and analysed in a timely manner so that suspicious activity is acted on.
Who Should Target It
ML3 suits organisations that expect to be specifically targeted by capable adversaries: agencies and suppliers handling highly sensitive information, operators of critical infrastructure, and large financial institutions. It needs mature asset management, change control and a security team able to respond to what the logs show.
How to Test for It
- Emulate a specific threat actor relevant to your sector, using techniques chosen to evade the controls you have deployed. See adversary emulation vs simulation.
- Attempt application control bypasses on servers, including vulnerable or unsigned drivers.
- Attempt credential theft and reuse against privileged accounts to test just-in-time administration and credential protection.
- Attempt to defeat phishing-resistant MFA through session theft, device registration abuse and help desk social engineering.
- With administrator-level access in a controlled exercise, attempt to delete or encrypt backups.
- Measure whether the security team detects and responds to each step. A purple teaming exercise is an efficient way to tune detection.
Our guide to Essential Eight penetration testing explains how these tests map to each strategy, and red teaming is the natural way to test ML3 controls together.
How to Reach and Keep ML3
Reach ML2 across all eight strategies first. Then invest in the processes ML3 depends on: fast, reliable patch pipelines, server application control with change management, just-in-time privileged access and a monitored logging platform. A maturity level assessment shows which ML3 requirements are still unmet.
Regulatory and Framework Context
ML3 is the top level of the Essential Eight Maturity Model. For an overview of the strategies and who must comply, see ASD Essential Eight explained. ASD revises the model periodically, so always check the current version on cyber.gov.au.
Latest intel on this topic
- ASD Essentials: evolving our approach to cyber security resilienceASD's ACSC · 7 October 2026
- Consultation on evolution of Essential EightASD's ACSC · 2 September 2026
Updated hourly from the daily briefing.
Common questions
Does ML3 make an organisation secure against nation-state attackers?
No framework guarantees that. ML3 makes compromise much harder and slower, and makes it far more likely to be detected, but a determined and well-resourced adversary may still succeed. ASD positions the Essential Eight as a baseline, not a complete security program.
Should every organisation aim for ML3?
Not necessarily. ML3 demands significant ongoing effort. It suits organisations facing targeted threats, such as those holding highly sensitive information or operating critical services. Most should reach ML2 consistently across all eight strategies first.
How is ML3 validated?
Through direct testing of each control plus realistic adversary exercises. At ML3 the question is less whether controls are configured and more whether they and the people monitoring them stop a skilled attacker who is trying to evade them.
Change log
- 13 November 2026First published.