Skip to content
fixGuide

Parked domain DMARC p=reject: records for domains that never send

Domains that never send mail should still publish DMARC, usually p=reject, plus SPF v=spf1 -all and a null MX. Here is the record set and why p=none is the wrong default.

You searched for

“parked domain dmarc p=reject”

Last verified: Oct 21, 2026Published: Oct 21, 2026

A domain that never sends mail is still a From address. Parked pages, leftover product names, unused country-code domains, and names you registered so someone else could not all work the same way in SMTP: anyone can put that name in the header. Inbox providers will look up SPF and DMARC on it. If those lookups find nothing, there is no policy telling them to refuse the forgery.

On a sending domain you start at p=none and tighten after you see real sources. On a no-send domain there are no real sources. The useful policy is p=reject from the start.

For how SPF, DKIM, and DMARC split the work, see SPF vs DKIM vs DMARC. The alignment rules that matter when mail actually goes out are in DMARC alignment explained. This page is the other case: the name should stay silent.

What counts as a no-send domain

Treat a name as no-send when nothing legitimate uses it in From or as an envelope domain:

  • A parked or “coming soon” site with no mailboxes.
  • A brand or product you retired, including an expired-looking public name that still resolves.
  • A country-code or extra TLD you hold for the same company (example.net sitting idle while example.com sends).
  • A defensive registration: you bought it so a competitor could not.

The test is traffic, not the registrar’s “parked” flag. If a forgotten CRM, billing tool, or forwarding address still sends as that hostname, it is a sending domain. Inventory senders first. If you are unsure, publish p=none with rua= for a short window, read the reports, then move to reject when the only rows are spoofing. Do not leave that monitor policy in place as the permanent setup.

The record set

Publish these on the no-send hostname. Replace unused.example with the name you are locking down.

SPF (TXT on the domain):

v=spf1 -all

-all means no IP is authorized. Do not copy include: mechanisms from the domain that actually sends. A leftover ESP include on a parked name authorizes that ESP to pass SPF for a brand you are not using.

Null MX (MX on the domain):

0 .

That is RFC 7505’s null MX: preference 0, exchange .. It tells receivers this name does not accept mail, which is the honest statement for a parked domain. Remove ordinary MX hosts that still point at a provider you no longer use, or mail directed at the parked name can still be delivered somewhere.

DMARC (TXT on _dmarc.unused.example):

v=DMARC1; p=reject; sp=reject;

Optional reporting:

v=DMARC1; p=reject; sp=reject; rua=mailto:[email protected]

p=reject asks receivers to refuse unauthenticated mail that uses this From domain. sp=reject says the same for subdomains that have no record of their own. rua= is how you watch spoofing attempts; it does not change the policy.

One TXT record per job. Do not publish a second v=spf1 or a second v=DMARC1 on the same name.

Skip DKIM unless something signs

DKIM needs a selector (selector._domainkey.domain) and a sending system that holds the matching private key. A parked domain has neither. There is no selector to publish, and a placeholder DKIM TXT does not make spoofed mail fail. SPF -all plus DMARC p=reject is the protection.

If you find old DKIM records from a tool you turned off, you can delete them. Leaving them does not hurt DMARC reject, but they are unused keys sitting in DNS.

Do not park at p=none

p=none is a monitor policy. Receivers still accept mail that fails authentication; they may send you XML about it. That is the right first step on a domain with payroll, marketing, and ticket mail you have not mapped yet.

It is the wrong steady state on a name with zero legitimate sources. Every unauthenticated message that claims that From can still be delivered, because you asked for reports instead of refusal. If reports already show no real senders (or you never stood up mail on the name), move to p=reject. Waiting for a round of “just in case” monitoring that never ends is how unused ccTLDs stay easy to spoof.

If a leftover sender appears after you reject, that mail will bounce, which is the signal you missed a tool. Put that tool on a domain that is supposed to send, with aligned SPF and DKIM, rather than opening the parked name back up.

Subdomains and sp=

DMARC on the organizational domain covers subdomains that do not publish _dmarc. sp= is the policy used for those names. Setting sp=reject next to p=reject makes the no-send rule explicit for www., shop., random hostnames, and typosquatting-style subdomains that only exist in a phishing From line.

If a subdomain does send (for example mail.example.com while example.net is parked), that sending hostname needs its own SPF, DKIM, and _dmarc record. A parent sp=reject will otherwise ask receivers to refuse it. Split the identities: lock the silent name, authenticate the one that mails.

Relaxed alignment can also let a subdomain DKIM or SPF pass count for a parent From. That is why you do not want a forgotten subdomain still signing as a parked parent. Details are in DMARC alignment explained.

Reports are optional, not the lock

p=reject works without rua=. Add reporting when you want to see who tried. Aggregate rua is enough; skip ruf= unless you have a reason to collect failure samples.

If rua= points at a domain you do not control, complete external destination authorization or many receivers will not send the files. Email Watch can ingest that rua stream so you are not opening XML by hand. It does not send or block mail; receivers apply the policy you publish. For a single parked name, the free checker is enough to confirm the records exist.

Check the live records

  1. Confirm nothing legitimate still sends as the hostname (mailbox, ESP, CRM, forwarding).
  2. Publish SPF v=spf1 -all, the null MX, and _dmarc with p=reject; sp=reject;.
  3. Wait for TTL.
  4. Run the free DMARC checker on that exact name. Confirm policy is reject, SPF is deny-all, and you did not leave a second SPF or DMARC TXT.

If you later want that name to send, change these records before the first campaign: real SPF includes, DKIM for the sender, and a DMARC policy you can live with while you watch reports. A reject record with -all will not deliver new product mail.

When a no-send name is one of many in a portfolio and you want rua on a mailbox you already read, start a trial and point reporting there. The records above are what lock the domain; monitoring only tells you who knocked.

Confirm the fix

Run the free check to confirm

Applied the fix above? Run the freeDMARC Checkeragainst your domain and see your grade update in real time.

Run the free check: DMARC Checker

Common follow-up questions

Should a parked domain publish DMARC?

Yes. Receivers look up _dmarc on the From domain of every message. If that lookup finds nothing, anyone can put your unused name in From and the policy decision never happens. Publish p=reject (and usually sp=reject) so forgeries of that name are refused.

What DNS records should a domain that never sends mail have?

SPF TXT v=spf1 -all (nobody is authorized to send), a null MX of 0 . (this name does not accept mail), and DMARC TXT on _dmarc with v=DMARC1; p=reject; sp=reject;. Add rua= if you want aggregate reports of spoofing attempts. Skip DKIM unless something actually signs mail.

Do I need DKIM on a parked or unused domain?

No. DKIM is a public key for a selector that a sending system uses to sign mail. If nothing sends, there is no selector to publish and no key to rotate. An empty or leftover DKIM record does not protect the domain; DMARC p=reject and SPF -all do.

Why not leave a no-send domain at p=none?

p=none only asks for reports. Receivers still deliver unauthenticated mail that claims the domain. On a name with zero legitimate sources there is nothing to learn by watching, and every day at none is a day spoofed mail can land. Move that domain to p=reject once you are sure nothing real sends as it.

What does sp=reject do on a parked domain?

sp= is the policy for subdomains that do not publish their own DMARC record. If you omit it, receivers apply p= to those subdomains anyway. Setting sp=reject makes the no-send stance explicit for names like shop.unused-brand.com. If a subdomain does send, give it its own SPF, DKIM, and DMARC instead of inheriting reject.

Should I add rua= on a domain that never sends?

Optional. rua= is how you see who tried to send as the domain. It is useful if you care about spoofing volume on a defensive registration. It is not required for p=reject to work. If rua points at a third-party domain (including a monitor), that destination must authorize external reports or many receivers will not send them.

We bought a ccTLD so a competitor could not. Does it need the same records?

Yes, if it never sends. Defensive names, unused country-code domains, expired product brands, and parked pages are all From identities someone else can type. The record set is the same: deny-all SPF, null MX, DMARC p=reject. Do not copy the sending domain's includes onto a name that should stay silent.

How do I confirm the parked-domain records after publishing?

Wait for DNS TTL, then run the free DMARC checker on the exact hostname. Confirm SPF is a single v=spf1 -all, MX is the null record (or no mail hosts you still operate), and _dmarc shows p=reject. If you added rua, send a test only after you change these records, because a reject policy will refuse unauthenticated mail from that domain.