Skip to content
Infrastructure

Reverse DNS and PTR records for email

Reverse DNS maps a sending IP back to a hostname, and receivers check it before anything else. Forward-confirmed rDNS, generic PTR problems, and how to fix them.

Reverse DNS answers the mirror of the usual question: given an IP address, what hostname does it belong to? For email it is a first-impression check that runs before any content or authentication is examined: the receiving server looks up the connecting IP's PTR record, and what it finds, a meaningful mail hostname, a generic ISP label, or nothing, shapes how the rest of the conversation is treated. Sending IPs without proper reverse DNS get scored down or refused outright, and the fix is one record plus one matching forward entry.

How the check works

Forward-confirmed rDNS, verified by hand
1# 1. Reverse: what does the IP claim to be?
2dig -x 198.51.100.42 +short
3# out1.mail.example.com.
4
5# 2. Forward: does that name point back?
6dig A out1.mail.example.com +short
7# 198.51.100.42
8
9# Match in both directions = forward-confirmed (FCrDNS)

The two-step matters because a PTR alone is trivially claimable by whoever controls the address space; the forward confirmation proves the domain owner agrees. Receivers run exactly this pair on every inbound connection, and Gmail's sender guidelines state the requirement plainly: sending IPs must have valid reverse DNS that matches a forward lookup. Beyond passing the check, the hostname itself communicates: out1.mail.example.com announces deliberate mail infrastructure, while a generic label like 198-51-100-42.dynamic.isp.net announces a residential connection, which is the profile of botnet spam and is filtered accordingly.

Who sets it, and how

PTR records live in the address owner's zone, not yours: the hosting provider, cloud platform, or ISP that allocated the IP controls the in-addr.arpa entry. Every serious platform exposes it, as a console field (Hetzner, OVH), an API setting (AWS via request or resource attribute, depending on service), or a support request. The forward half, the A record for the chosen hostname, lives in your own DNS and is yours to publish. Set both when infrastructure is provisioned, and re-verify whenever IPs change, since the PTR does not migrate with your workload.

Practice and pitfalls

Name the host meaningfully under your organizational domain, and let the MTA announce the same name in its HELO/EHLO greeting: PTR, forward record, and HELO agreeing is the coherent identity receivers want, and some run the comparison explicitly. One PTR per sending IP is the norm; multiple PTRs confuse more than they help. The recurring pitfalls: cloud IPs recycled with a previous tenant's PTR still attached, IPv6 sending paths without PTR coverage while receivers hold IPv6 to the stricter standard (Gmail rejects on it), and infrastructure moves where the new IPs simply never got records because the checklist lived in someone's head.

Frequently asked questions

Does reverse DNS affect deliverability directly?
Yes, at the connection layer: missing or generic rDNS costs acceptance and scoring at most receivers, and IPv6 without valid rDNS is rejected outright at Gmail. It is a pass/fail hygiene item rather than a ranking factor to optimize.
What hostname should the PTR use?
A descriptive name under your domain that resolves back to the IP: out1.mail.example.com, mta-3.example.com. Avoid names on the provider's domain where custom rDNS is available, since the hostname is part of your identity.
My provider will not let me set a PTR. Now what?
Then the IP is unsuitable for direct sending; this is common on residential and small-VPS ranges, many of which sit on blocklists preventively. Relay through a smarthost or ESP whose infrastructure has proper rDNS instead.
Is rDNS checked for the domain or per IP?
Per connecting IP. Every address that opens SMTP connections to the world needs its own forward-confirmed PTR, which is why the audit belongs in infrastructure provisioning rather than DNS housekeeping.

Key takeaways

  • Receivers check the connecting IP's PTR before anything else, and forward confirmation is the standard
  • Generic or missing rDNS reads as botnet territory; meaningful hostnames under your domain read as infrastructure
  • The PTR lives with whoever owns the IP; the matching forward record lives with you, and HELO should agree with both
  • IPv6 sending without valid rDNS fails outright at Gmail, and recycled cloud IPs arrive with stale PTRs
  • ESP shared IPs handle this for you; dedicated IPs and self-hosted MTAs make it your provisioning checklist item