How Many Domain Names Exist, and Why Every Count Differs
Industry reports put the world at about 412 million domain registrations at the end of the third quarter of 2026, but that number depends on what you count and who supplies the data. Here is how the figures are built and where they diverge.
The headline figure and its source
The most widely cited total comes from the Domain Name Industry Brief (DNIB), a quarterly report sponsored by Verisign, the operator of the .com and .net registries. For the end of the third quarter of 2026, DNIB.com reported 411.6 million domain name registrations across all TLDs. Verisign announced it on 22 October 2026. That was up 9.9 million (2.5%) from the second quarter and 33.0 million (8.7%) from a year earlier. For comparison, the third quarter of 2025 stood at 378.5 million.
Verisign's announcement of the Q3 2026 results put .com at 169.8 million and .net at 12.6 million (182.5 million together), figures we saw only in trade-press coverage of the release, not in the release itself. The table below stays on the second quarter of 2026, where every segment comes from the same Verisign release, so the rows are comparable:
| Segment (end of Q2 2026) | Registrations | Note |
|---|---|---|
| All TLDs | 401.6 million | First time DNIB crossed 400 million |
| .com | 166.6 million | Roughly 41% of the total |
| .net | 12.5 million | .com and .net together are about 179.1 million |
| All country-code TLDs | 148.6 million | Largest by registrations: .cn, .de, .uk, .ru, .nl, .br, .fr, .au, .in, .eu |
Subtracting those segments from the total leaves roughly 74 million registrations in everything else: .org, .info, .biz and the other legacy gTLDs, plus the hundreds of new gTLDs. That remainder is a derived number, not a published one. Treat it as an approximation.
Always state the quarter with a count. "400 million domains" without a date is meaningless, since the base grows by several million a quarter, and the number you read may be a year old.
What exactly is being counted
A "domain registration" in these reports is a registered name at the level the registry sells, usually a second-level label such as example.com. It is not a website, a hostname, or a company. That definition creates several conventions you should know before comparing any two numbers.
Registrations, not unique brands
If one company registers brand.com, brand.net, brand.org and brand.shop, the total goes up by four. The label "brand" is counted once per TLD. The number of distinct labels is therefore lower than the number of registrations, and nobody publishes a reliable global figure for it.
Registrations, not active sites
A large share of registered names show no meaningful website. Some are parked, some redirect, some exist only for email, defensive registration or speculation. Counting registered names says nothing about how many are in use. This is why you should never turn a registration count into "number of websites".
Second-level versus third-level
In ccTLDs that sell below a public second level, such as example.co.uk or example.com.br, whether a report counts the name at the second or third level changes the total. Different reporters make different choices, and not all say which.
Three counting methods and why they give different numbers
1. Registry reports
Registries know their own database. Verisign publishes .com and .net numbers from its own systems, other registries report to ICANN monthly, and aggregators such as DNIB compile them. This is the most authoritative source for gTLDs. It counts every registered name, including those on hold or in the grace period, so it runs higher than any zone-file count of the same TLD.
2. Zone-file counts
A zone file lists names that currently have nameserver delegations in a TLD. Counting unique owner names gives a delegated-domain count. It is independent of the registry's own reporting, which makes it a good cross-check, and it is the method behind services built on ICANN CZDS zone files. But it excludes any registered name that is not published:
- names on
clientHoldorserverHold; - names registered without nameservers;
- names in the pending-delete stage at some registries;
- names that were registered and removed within the same day, so they never appear in a daily snapshot.
For a typical TLD the zone count is therefore a little lower than the registry's registration count. How much lower varies by TLD and over time.
3. Registrar counts
A registrar can say how many domains it manages, usually as a headline marketing number. That is the number of registrations under management at one company, across many TLDs. Registrar counts overlap with nothing and sum to roughly the registry total only if you can see every registrar, which no public source does. They are good for ranking registrars, bad for sizing the market. Transfers between registrars can also briefly appear in both queues.
| Method | Strength | Blind spot |
|---|---|---|
| Registry reports | Authoritative for that TLD | ccTLD reporting is voluntary and uneven |
| Zone-file counts | Independent, daily, per-name detail | Misses undelegated names; not all TLDs are published |
| Registrar counts | Direct view of customer holdings | One company only; marketing definitions |
The ccTLD gap
Verisign-sponsored reporting has a clear data advantage for .com and .net and for TLDs whose contracts with ICANN require published reporting. Country-code operators are not bound by ICANN's registry agreement, so each decides what to publish. Some release monthly counts, some release occasionally, some publish nothing. Aggregated ccTLD totals therefore combine reported figures and estimates, and DNIB's reports have historically noted where numbers are estimated; check the report's footnotes for the quarter you cite.
Zone files do not fill the gap on their own, because most ccTLDs do not distribute their zones through CZDS. If a service says it counts a given ccTLD, ask where the data comes from. Counts for some ccTLDs are built from passive DNS, web crawls or operator statistics rather than from a zone. The contrast between the two families is explained in gTLD vs ccTLD vs new gTLD.
Even the biggest ccTLD figures carry caveats, and one reason a ccTLD ranking can look different in different sources is that each operator defines and counts its own registrations.
What gets double-counted or miscounted
- Same label across TLDs. Counted once per TLD, as above.
- Transfers in progress. A name moving registrars is still one registry object, but registrar-level counts may show it twice during the transfer window.
- Grace and redemption names. Registry counts include expired names still held in the lifecycle, so the figure includes domains that are on their way out. The domain lifecycle can last well over a month in some TLDs after expiry.
- Internationalized domain names. Each IDN is stored as a punycode string, which appears once. Counting both the Unicode and ASCII forms would double it; reputable reports do not.
- Subdomains. Billions of hostnames exist beneath registered domains. They are not domain registrations and must not be added to these totals.
How zone-file-derived counts are built, in general
Services in the style of TLD analytics sites generally follow the same pipeline. They download each participating TLD's zone file once a day, extract the unique owner names that carry NS records, and store the set. Counting the set gives the delegated size of each TLD that day. Comparing today's set with yesterday's gives names that appeared (new registrations or names that regained delegation) and names that disappeared (deletions, holds or removals). Summing across TLDs yields an aggregate for the TLDs covered.
Three limitations follow directly. A TLD that does not publish its zone is invisible. A name that appears is not necessarily new, because a name restored from hold re-enters the zone; this is why a newly registered domains list carries some noise. And a name that disappears is not necessarily deleted. None of this is a flaw in zone-based counting, only the definition of what it measures. TLDlive describes its own lists in these terms: delegation changes in zone files, not registration facts.
Which number should you use?
| Your question | Best source |
|---|---|
| How big is the whole namespace? | The latest DNIB figure, with its quarter |
| How big is .com, .net, or a gTLD? | The registry's own report |
| How fast is one TLD growing day to day? | Zone-file delta counts |
| How does a ccTLD compare? | The ccTLD operator's published statistics, with the caveat that definitions vary |
| How many sites exist? | None of the above; use crawl-based web surveys |
For research or articles, quote the source and the period together, for example "Verisign's DNIB, end of Q3 2026", and avoid summing figures from different publishers.
Sources
- Verisign blog: Q3 2026 Domain Name Industry Brief (411.6 million, published 22 October 2026)
- Verisign investor release: DNIB.com reports 411.6 million registrations, Q3 2026
- Verisign investor release: 401.6 million registrations, Q2 2026 (.com 166.6 million, .net 12.5 million, ccTLDs 148.6 million)
- Verisign investor release: 378.5 million registrations, Q3 2025
- Trade-press summary of the Q3 2026 .com and .net figures (webhosting.today)
Frequently asked questions
How many domain names are registered worldwide?
Verisign's Domain Name Industry Brief reported 411.6 million registrations across all TLDs at the end of the third quarter of 2026. The figure is quarter-end and counts registrations, not websites.
How much of that is .com?
At the end of the third quarter of 2026, trade-press coverage of Verisign's results put .com at 169.8 million registrations and .net at 12.6 million. That makes .com about 41% of the 411.6 million total. The second-quarter figures were 166.6 million and 12.5 million.
Why do zone-file counts differ from registry numbers?
Zone files list only names with active delegations. Names on hold, registered without nameservers, or otherwise not published are in the registry's database but absent from the zone, so the registry total is usually somewhat higher.
Are ccTLDs fully included in the totals?
Mostly, but with less precision. Country-code registries decide what to publish, so some figures are estimates and some TLDs report irregularly. Check the footnotes of the report you cite.
Does the total equal the number of websites?
No. Many registered names are parked, redirected, held for email only or unused, and billions of subdomains sit beneath registered names but are not registrations.
How often do the numbers change?
Verisign publishes quarterly, with net additions of roughly 9 to 10 million in each of the last two quarters. Zone-file based services update daily, but cover only the TLDs whose zones they can obtain.