Compliance and requirements
DMARC and UK government guidance: what the NCSC recommends
By DMARCHub Team5 min read
On this page
The National Cyber Security Centre (NCSC), the UK's technical authority on cyber security, recommends that every organisation deploys SPF, DKIM and DMARC, and moves DMARC to an enforcement policy of p=reject. For central government domains this is not advice but a requirement: gov.uk domains have been expected to enforce DMARC since 2016, and the same expectation increasingly reaches suppliers.
Who is the NCSC and why does its guidance matter?
The NCSC is part of GCHQ and acts as the UK's national authority for cyber security. Its published guidance is not law, but it sets the baseline that government departments, regulators, insurers and procurement teams treat as good practice. When a framework or contract asks whether you follow "recognised guidance" on email security, NCSC guidance is usually what they mean.
Its email security and anti-spoofing guidance is straightforwardly practical. It explains why an unauthenticated domain can be impersonated, and it names the three controls that fix that: SPF (Sender Policy Framework, RFC 7208), DKIM (DomainKeys Identified Mail, RFC 6376) and DMARC (Domain-based Message Authentication, Reporting and Conformance, RFC 7489). If those acronyms are new, start with what DMARC is and how the three protocols work together.
What does the NCSC actually recommend?
Condensed, the guidance for any sending domain is:
- Publish SPF listing every service authorised to send for the domain, and end the record with a hard fail (
-all) once the list is complete. - Sign with DKIM on every sending service, so messages carry a cryptographic signature tied to your domain.
- Publish DMARC, starting at
p=nonewith an aggregate reporting address (rua), so you can see who is sending as your domain before anything is blocked. - Move to enforcement: tighten to
p=quarantineand thenp=rejectonce the reports show legitimate mail authenticating correctly. - Protect domains that never send email with SPF
v=spf1 -all, an empty DKIM posture and DMARCp=reject, because parked domains are the easiest to spoof.
That last point is often missed. If your organisation holds brand-protection or legacy domains, the NCSC expects them locked down too; the approach is covered in DMARC for non-sending domains.
Beyond authentication, the same guidance family recommends TLS for mail in transit, including MTA-STS and TLS reporting. Those sit alongside DMARC rather than replacing it.
The 2016 government mandate: p=reject as the destination
In 2016 the UK government set a requirement for central government domains: gov.uk domains had to publish DMARC, with enforcement at p=reject as the destination rather than an aspiration. This made the UK one of the first governments to mandate DMARC at scale, and the visible result is that spoofing a well-run departmental domain directly is now largely a dead end for attackers.
The mandate matters outside government for two reasons. First, it proved the rollout model works: monitor at p=none, fix legitimate senders, then enforce. That is exactly the staged approach described in moving from p=none to p=reject, and it is the same path the NCSC recommends to businesses. Second, it shifted attacker attention to softer targets: suppliers, arm's-length bodies and lookalike domains, which is where the current pressure sits.
What was Mail Check, and what happened to it?
Mail Check is the NCSC's own platform for helping UK public sector organisations assess their email security. For years it did two jobs: it checked the configuration of SPF, DKIM, DMARC, MTA-STS and TLS records, and it received and analysed DMARC aggregate reports on behalf of enrolled organisations, acting in effect as a free DMARC report processor for the public sector.
In 2025 the NCSC refocused the service. Mail Check stopped processing DMARC aggregate reports and concentrated on configuration checking, with organisations directed to handle report analysis themselves or through a commercial DMARC reporting service. The change did not weaken the recommendation to collect and read reports; it moved the responsibility for doing so back to each organisation.
For anyone who relied on Mail Check for reporting, the practical takeaway is that the rua address in your DMARC record needs to point somewhere that still parses the XML. The reports themselves are unchanged: counts, sending IPs and authentication results per source, as explained in how to read DMARC reports.
What does this mean for suppliers to the public sector?
If you sell to government, the NCSC baseline tends to arrive via procurement. Security questionnaires and frameworks increasingly ask for your DMARC policy explicitly, and "we have a record at p=none" reads very differently from "we enforce p=reject". Some buyers treat enforcement as a pass-fail item; others score it.
There is also a fraud dimension. A supplier domain without enforcement can be spoofed to send fake invoices or payment-change requests to the very department it serves, and the department's own p=reject record does nothing to stop that. The mechanics of that attack, and why your customers bear the cost of your unprotected domain, are covered in supplier email fraud.
A sensible supplier posture, aligned with the guidance, looks like this:
- Publish DMARC at
p=nonewith aggregate reporting on every domain you own, including non-senders. - Use a monitoring period to identify every legitimate sending service and fix SPF alignment and DKIM signing for each.
- Move senders to
p=quarantine, thenp=reject, and lock non-sending domains atp=rejectimmediately. - Keep collecting reports after enforcement, because new services and misconfigurations appear over time.
The NCSC's guidance itself is published openly at ncsc.gov.uk and is worth reading first-hand; it is unusually free of vendor framing. Reading the aggregate reports it tells you to collect is the tedious part, and that is where a monitoring service such as DMARCHub turns the raw XML into readable dashboards; see /features if you want that job done for you.
Frequently asked questions
Is NCSC DMARC guidance legally binding for private companies?
No. For private organisations it is guidance, not regulation. It becomes contractually relevant when a public sector buyer, a framework agreement or a cyber insurance policy references it as the expected standard, which is increasingly common. Government domains are a different case: for them, DMARC enforcement is a mandated requirement.
Does the NCSC recommend p=quarantine or p=reject?
The end state is p=reject. The guidance treats p=quarantine as a useful intermediate step while you build confidence in your reports, not as a destination. A domain that sits at quarantine indefinitely still lets some spoofed mail reach junk folders, where users can retrieve it.
Can I still use Mail Check for DMARC reporting?
No. Mail Check remains available to eligible UK public sector organisations for configuration checking, but it no longer processes DMARC aggregate reports. Organisations that used it as their rua destination need to point their records at another report processor and re-establish their own visibility.
Does following the NCSC guidance help with Cyber Essentials?
Cyber Essentials does not currently test DMARC directly, so you can pass without it. But the two come from the same stable, and assessors, insurers and buyers who see Cyber Essentials will often ask about email authentication in the same breath. Deploying DMARC alongside Cyber Essentials controls closes the spoofing gap the certification alone leaves open.