How were the 58 resolvers actually tested?

Each was queried for A, MX, TXT and NS records on ten real domains, using the backend's actual 5-second timeout and single retry — the same conditions a real user experiences, not a synthetic lab test.

Why does a resolver get dropped even if it usually works?

A resolver that fails 10-45% of queries looks fine in casual, occasional use but causes real, hard-to-diagnose failures at scale — reliable enough to seem fine, unreliable enough to matter.

Does this list ever change?

Yes — it's re-verified against live behavior rather than assumed to stay accurate. Several previously-used resolvers (Controld, FreeDNS, 101domain, CNNIC, BTTC) had simply stopped answering DNS at all by the time they were rechecked.

We Tested 58 Public DNS Resolvers — Here's What Actually Answers

Most "best public DNS resolver" lists online are years old, or copied from each other without anyone re-checking. Ours comes from actually querying 58 of them and writing down what happened.

Why Measure Instead of Trust a List

A resolver that worked reliably when a ranking was written can quietly stop answering, get renumbered, or start dropping a fraction of queries — and a static list has no way to notice. The only way to know which resolvers currently work is to ask them.

How We Tested

58 public resolvers were queried for A, MX, TXT and NS records on ten real domains (apple.com, google.com, github.com, and others), using the same 5-second timeout and single retry our backend actually applies to every request — the conditions a real user experiences, not a synthetic best case.

The 18 That Made the Cut

Google, Cloudflare, Quad9 and Quad9 Alt, OpenDNS, Dyn, NextDNS, AdGuard, DNS.SB, CleanBrowsing, SafeDNS, Yandex, 114DNS and 114DNS Alt, AliDNS and AliDNS Alt, HiNet, and KT. Verified against the real backend, all 18 answered for every single domain tried.

The 5 That Were Cut

Verisign, Neustar, Baidu, DNSPod and CNNIC were dropped after losing 10 to 45 percent of queries in testing — reliable enough to look fine in casual use, unreliable enough to cause real, hard-to-diagnose failures at scale.

Why This List Isn't Permanent

Resolver reliability changes over time — some of the resolvers this exact list replaced (Controld, FreeDNS, 101domain, CNNIC, BTTC among them) had simply stopped answering DNS at all by the time we checked. The list gets re-verified against live behavior rather than assumed to still be accurate.

What This Means for You

If you're choosing a resolver to configure on a router or device, this reflects what's actually been observed to answer reliably, not a marketing claim or a list nobody has rechecked in years.

See These Resolvers in Action

Run a propagation check →Next: Why DNS checks fail from Russia, China and Taiwan →

Try it on your domain

See the theory in practice — run a DNS Propagation Checker for your own domain.