The legal basis: Article 21(2)(i) and (j)
NIS2 does not contain a standalone access control article. The requirement sits inside Article 21's list of risk-management measures, alongside asset management under Article 21(2)(g) and the human-resources and access aspects of basic cyber hygiene. Access control is therefore judged as part of the whole control set: proportionate to risk, documented, and operating in practice.
For the provider types listed in Article 1 of Commission Implementing Regulation (EU) 2024/2690, the annex to that regulation expands the Article 21 measures into more detailed technical and methodological requirements, including access control topics such as privileged access and authentication. Entities in those categories should design against the annex as well as the Directive text.
Authorise access from roles and responsibilities
Access should start from a decision, not from a request for convenience. Each system should have an owner who approves access, and permissions should be granted against defined roles that match job responsibilities. Individual one-off permissions accumulate quickly and are where reviews later fail, so exceptions should be recorded with a reason and an expiry.
Least privilege is the working principle: accounts receive the access needed for their duties and no more, and standing administrative rights are replaced by time-bound elevation wherever the tooling allows. Service and application accounts belong in the same model, with named owners and secrets managed like other credentials.
- Named system owners approve and periodically re-confirm access
- Role-based permissions aligned to job responsibilities
- Time-bound elevation instead of permanent administrative rights
- Exceptions recorded with a reason, an owner and an expiry date
Privileged access deserves its own controls
Administrative accounts are the highest-value target in most intrusions, and access control policies should treat them as a separate class: separate accounts for administrative work, stronger authentication, restricted source networks or jump hosts where proportionate, and logging that is reviewed rather than merely collected.
Multi-factor authentication is named in Article 21(2)(j) as a measure to use where appropriate; for remote and privileged access it is difficult to argue it is not appropriate. Where legacy systems cannot support it, the compensating controls and the replacement plan should be documented as a risk decision.
Joiners, movers and leavers
Access control fails most often at change points. Role transfers should remove permissions attached to the old role, not simply add the new ones, and departures should disable accounts within a defined, short period - immediately for privileged accounts and for anyone leaving under adverse circumstances. Human resources processes and IT account management need a shared trigger so that no departure depends on someone remembering to send an email.
Supplier and temporary accounts are part of the same lifecycle. External access should expire by default, be sponsored by a named internal owner, and be reviewed while it is active rather than discovered after the engagement ends.
Review access and test that removal works
Periodic access reviews ask system owners to confirm that each account and permission is still needed. Reviews should produce evidence: the list reviewed, decisions taken, removals made and exceptions accepted. A review that never removes anything is a signal it is not really examining the data.
Removal should also be tested directly. Sampling departed users and transferred staff against live account lists, and testing that disabled accounts cannot authenticate to connected systems, catches the synchronisation failures that reviews on paper miss. Findings feed the same remediation tracking as other Article 21 control tests.
Evidence and supervision
The evidence set is compact: the access control policy, role and permission definitions, approval records, privileged access arrangements, review outputs with remediation, and joiner-mover-leaver timeliness data. ENISA's NIS2 Technical Implementation Guidance discusses access control among the measures entities should be able to evidence, and national competent authorities can ask for these records during supervision or after an incident.
Handled this way, access control supports the rest of the NIS2 programme: incident handling depends on trustworthy identity records, and management oversight under Article 20 depends on being able to show that a named control actually operates.
Points of clarification
Does NIS2 require multi-factor authentication?
Article 21(2)(j) lists the use of multi-factor or continuous authentication solutions among the measures, qualified by appropriateness. For remote access and privileged accounts, MFA is the expected norm; where it genuinely cannot be applied, the entity should document the compensating controls and treat the gap as a recorded risk decision.
How often should access rights be reviewed?
The Directive does not set an interval. Reviews should be scheduled in the entity's own policy according to risk - privileged access more frequently than standard access - and triggered additionally by role changes, departures, incidents and organisational change.
Do access control requirements cover suppliers and temporary accounts?
Yes, in effect. Article 21 measures apply to the entity's network and information systems, and access granted to suppliers or temporary staff is access to those systems. Such accounts should be sponsored, time-limited, logged and reviewed like any other account, and removed when the engagement ends.