Persistence — MITRE ATT&CK® Tactic TA0003

Tactical objective

Adversaries preserve access through accounts, services, scheduled mechanisms, startup locations, application components, cloud resources, or configuration changes.

In ATT&CK, a tactic describes why an adversary performs an action. Persistence is not necessarily a single chronological phase, and techniques associated with it can appear repeatedly or alongside other objectives during one operation.

This BMC path highlights 5 high-value profiles: External Remote Services (T1133), Scheduled Task or Job (T1053), Boot or Logon Autostart Execution (T1547), Create Account (T1136), Server Software Component (T1505). The complete ATT&CK catalog contains additional techniques and sub-techniques, so this page is a teaching sequence rather than an exhaustive replacement for the official matrix.

What defenders should ask

The best tactical questions connect observable behavior to an asset, identity, trust decision, and consequence. They help investigators test competing explanations instead of inferring intent solely from an alert label.

• What change allows access to survive a restart or credential reset?
• Which identity or process created the persistence mechanism?
• Does the mechanism appear in an approved configuration baseline?
• Which part of the environment is outside current telemetry or control scope?
• What earlier and later behaviors would strengthen or weaken this hypothesis?

Architecture and preventive controls

Prevention of persistence behavior rarely depends on one product. Architecture, identity, configuration, network paths, software maintenance, user workflow, and recovery design all influence the available opportunities.

Control design should identify where the organization can deny, constrain, slow, or expose the behavior. For every control, define the expected evidence, exception process, failure mode, and owner.

• Restrict who can create accounts, services, tasks, and startup entries.
• Monitor configuration drift and integrity of application components.
• Use privileged workflows for persistence-capable changes.
• Review cloud identities, keys, applications, and automation regularly.

Detection and evidence

Detection for persistence should focus on behavior and context rather than an identifier alone. The same command, protocol, or configuration action may be authorized or malicious depending on the actor, asset, sequence, and expected workflow.

Use multiple evidence sources when possible. Independent identity, endpoint, network, cloud, and application records are especially valuable when an adversary can alter host-local evidence.

• Account and credential lifecycle events
• Service, task, startup, and application-component changes
• Cloud identity and resource creation
• File-integrity and configuration-baseline alerts
• Asset criticality, identity role, and change-window context
• Correlated preceding and follow-on behaviors across tactics

Investigation and response

Build a timeline that separates observed actions from inferred objectives. Preserve the event, process, account, resource, destination, and configuration evidence supporting every mapping.

Response priorities for persistence should address the active behavior while also identifying the access and dependencies that made it possible. Containment that removes one symptom but leaves the enabling identity, trust path, or persistence mechanism can create a false sense of closure.

• Confirm the affected scope and earliest supported activity.
• Identify accounts, hosts, workloads, applications, and external infrastructure involved.
• Preserve volatile and centralized evidence before making disruptive changes.
• Contain the enabling path and validate that controls are functioning.
• Search for related techniques and repeat activity across the environment.
• Document confidence, residual risk, and the conditions for recovery.

Safe tabletop exercise

Create a fictional scenario in which an adversary pursues the Persistence objective against a representative business service. Select two techniques from this path and identify the architecture, identity, and evidence required for the behavior.

Have one participant act as the investigator and another as the service owner. Compare what security tooling reports with what the service owner knows about expected behavior. Record one prevention gap, one detection gap, and one recovery or resilience improvement.

• State the assumed adversary access and objective.
• Define the affected platform and business service.
• List expected benign activity that could resemble the behavior.
• Identify the minimum evidence needed to reach a defensible conclusion.
• Map only what the scenario actually supports.

Attribution and scope

© 2026 The MITRE Corporation. This work is reproduced and distributed with the permission of The MITRE Corporation.

MITRE ATT&CK® and ATT&CK® are registered trademarks of The MITRE Corporation. Bare Metal Cyber is not affiliated with or endorsed by MITRE.

This article is an original Bare Metal Cyber educational treatment of tactic TA0003 reviewed against Enterprise ATT&CK v19.2. Verify current technique relationships and object status on the official tactic page.

Certification relevance

This subject appears in or supports the following certification bodies of knowledge:

Authoritative sources

• MITRE ATT&CK® Persistence Tactic
• MITRE ATT&CK® Data & Tools
• MITRE ATT&CK® Legal and Branding Guidance