To use a public NTP server list, filter the hostnames by provider, review each provider's leap-second handling, copy the visible ntpd-style configuration lines, and verify reachability from the target system before relying on them. A well-curated public NTP server list contains the eleven official hostnames most readers actually need: the rotating 0, 1, 2, and 3.pool.ntp.org names from the NTP Pool Project, the time.google.com family from Google Public NTP, time.cloudflare.com from Cloudflare, and the time.nist.gov round-robin address from NIST. Each row typically records the provider, scope, leap-handling approach, and a short operational note so you can compare strategies at a glance. The list gives you copy-ready server lines such as server time.google.com iburst, but it cannot test whether those hostnames actually respond from your network because ordinary web pages cannot send UDP packets to port 123. Treat the list as a starting configuration that you adapt to your time client — whether that is ntpd, chrony, systemd-timesyncd, Windows Time, or the settings page on a router.
Most readers arrive at a public NTP server list because they need to configure a single machine, a small office network, or a lab device and do not want to memorize every regional pool name. The Public NTP Server List exists precisely for that workflow: it keeps the official hostnames in one place, lets you search by provider or feature, and emits clean configuration snippets you can paste into a real config file.
What the Public NTP Server List Actually Contains
The list is a static reference of eleven rows keyed by official hostnames, not by IP addresses. Each row records only information the provider has published: scope (global or regional), leap-handling behavior (none, smear, or UTC(NIST) traceable), and a short operational note such as "round-robin hostname" or "do not mix with non-smearing servers". Because the entries are hostnames rather than IPs, the underlying addresses can rotate without forcing a configuration edit on your side.
The four providers represented cover the overwhelming majority of public NTP usage:
- NTP Pool Project — volunteer-run, rotating 0.pool.ntp.org through 3.pool.ntp.org, no leap smear, recommended as hostnames rather than pinned IPs.
- Google Public NTP — time.google.com plus four numbered siblings, leap smear enabled, global scope.
- Cloudflare Time Service — time.cloudflare.com on Cloudflare's anycast network, uses the leap indicator without smearing.
- NIST Internet Time Service — time.nist.gov round-robin address, distributes UTC(NIST).
Filtering is case-insensitive: typing "google" narrows the rows to the Google entries, while typing "smear" surfaces every provider that documents smearing behavior. The copy action emits one ntpd-style server line per visible hostname, with the iburst keyword already attached so the first synchronization happens quickly after the daemon starts.
Provider Strategies and Leap-Second Handling
Leap seconds are the single biggest reason a public NTP server list cannot be used as a grab-bag of "any four servers will do". A leap second is an extra second inserted into UTC to keep civil time aligned with Earth's rotation, and different providers react to it differently. The table below summarizes what each provider does, so you can pick one coherent strategy.
| Provider | Hostname pattern | Leap handling | Operational note |
|---|---|---|---|
| NTP Pool Project | 0/1/2/3.pool.ntp.org | Uses the leap indicator (no smear) | Use hostnames; volunteers change underlying IPs |
| Google Public NTP | time.google.com and time1–4.google.com | Leap smear | Do not mix with non-smearing servers |
| Cloudflare Time Service | time.cloudflare.com | Uses the leap indicator (no smear) | Anycast, global |
| NIST Internet Time Service | time.nist.gov | Distributes UTC(NIST) | Use the round-robin hostname, not a pinned IP |
The practical rule, repeated by Google's published NTP documentation, is that you must not mix smeared servers with non-smeared ones in the same pool. If your configuration points to time.google.com and pool.ntp.org at the same time, your client may briefly disagree with itself during a leap-second event and behave unpredictably. Pick one coherent source strategy and follow that provider's published usage terms.
Picking Hostnames and Copying the Configuration Lines
The fastest path through the Public NTP Server List is to narrow the rows to one provider, copy the visible entries, and then adapt them to the format your time client expects. The steps below assume you already know whether you want a smearing source or a leap-indicator source.
- Open the Public NTP Server List and type a provider name (for example, "google" or "nist") into the search field to narrow the table. You can also type a feature such as "smear" or a specific hostname fragment.
- Read the short operational note on each remaining row. Confirm that the leap-handling column matches what every other source in your configuration will do.
- If you plan to mix sources, switch back to the unfiltered view and review the leap-handling column across all eleven rows so you do not accidentally combine smeared and non-smeared servers.
- Click the copy action in the tool. The copied text is one ntpd-style server line per visible hostname, with the iburst keyword already attached to speed up the first synchronization.
- Paste the lines into a scratch buffer, then translate them into your time client's syntax (see the next section). Do not paste ntpd-format lines verbatim into a chrony or Windows configuration.
- From the target system itself, run a real reachability test such as ntptime, chronyc tracking, or the vendor's diagnostic command, because the list cannot perform that test from your browser.
Adapting the Lines to ntpd, chrony, and systemd-timesyncd
The copied lines assume ntpd's syntax, where each server is declared on its own line with optional keywords such as iburst, prefer, or noselect. Most readers, however, run a different client. chrony uses server lines too, but expects an iburst keyword at the end and writes them to /etc/chrony.conf or /etc/chrony/chrony.conf. systemd-timesyncd reads /etc/systemd/timesyncd.conf and accepts a single NTP= line with space-separated hostnames; it does not understand per-line keywords such as iburst.
Windows Time, found on Windows 10 and 11 clients, is configured through Settings > Time & Language > Date & time > Additional clocks > Internet Time > Change settings, and accepts only one server hostname at a time. Network appliances and firewalls usually expose a single "primary NTP server" field in their admin UI, so you may have to choose the most important hostname rather than listing four. None of these targets accepts the raw ntpd lines from the list without editing; the output of the tool is a starting point, not a universal configuration.
Verifying Reachability from the Target System
A hostname that resolves and responds from one network may be blocked, rate-limited, or slow from another. Firewalls, captive portals, DNS policy, and provider maintenance all affect service. The browser in which you read the list cannot perform a real NTP check because ordinary web pages cannot send UDP packets to port 123, which is the port NTP uses. Run the reachability test on the actual machine you intend to synchronize.
On a Linux host running ntpd, ntpq -p lists the configured servers with a reachability column that climbs to 377 once a few rounds succeed. On chrony, chronyc sources -v reports a similar column. On Windows, w32tm /query /status shows the source name and the last successful sync. If a server never appears or stays at zero reach, check egress UDP 123 on your firewall and confirm that DNS resolves the hostname to a sensible address. If you need a defensive rule for that egress path, the related guide on using a common ports list as a firewall allowlist walks through the practical limits of such an approach.
When to Skip the Public List and Use Vendor Time Instead
For infrastructure where wrong time could affect authentication, logs, finance, safety, or incident response, do not rely on a single public hostname. Use the operating-system or equipment vendor's supported configuration, multiple monitored sources, and an authenticated or local time architecture such as a GPS-disciplined stratum-1 server. Active Directory domains expect member computers to follow the domain hierarchy; Kubernetes clusters usually expect chrony pointing at a local NTP pod; routers that double as VPN endpoints often need their own hardware time source for certificate validation.
A public NTP server list is the right tool for the long tail of cases — a developer laptop, a Raspberry Pi on a home network, a printer, a switch that needs basic time-of-day for log timestamps, or a lab device that just needs a clock close to real. For that work, copy the rows from the Public NTP Server List, adapt them to your client, and verify reachability once. The configuration that survives that verification is the one you keep.
Related reading: Flip Canvas in Photoshop: Keyboard Shortcut Setup.