Starlink latency is delay: how long a round trip takes. It is different from download speed and often explains why calls or games feel slow even when throughput looks healthy. This guide explains why the number moves and how to measure it. For the buying comparison, see Starlink versus fiber latency.
Latency is delay, throughput is volume
Throughput (the usual “Mbps”) is how much data can move in a second. Latency is how long you wait before a small message comes back.
A moving van and a postcard make the split obvious. The van is throughput: it can carry a lot. The postcard is latency: it answers a short question. You can have a large van and still wait on the postcard if the road is long or jumpy.
On a video call, your voice is a stream of small packets that must arrive soon. On a movie, the player can fill a buffer and hide delay. That is why a speed test can look “fine” while a call feels late.
Jitter is how much latency changes from packet to packet. Packet loss is data that never arrives. The three travel together in real time, but they are not the same. TNET shows ping publicly and retains jitter with saved results; packet loss must be checked in the Starlink app or the application reporting the problem. How to read speed test results explains the split, and loss has its own guide: elevated packet loss.
Where delay comes from on Starlink
Every internet path stacks waits: your device, Wi-Fi, the router, the dish, the space segment, the ground network, and the site you are talking to.
Starlink uses low-Earth-orbit satellites, so the space leg is much shorter than older geostationary satellite internet. Public user reports and third-party tests often describe latency that supports video calls and many games, while remaining higher and more variable than a good fiber circuit. Results still vary by location, routing, Wi-Fi and load, which makes a local measurement more useful than one universal millisecond figure.
Other ordinary waits still apply:
- A phone at the far end of the house.
- A VPN concentrator on another continent.
- A game server or meeting region far from the ground path you landed on.
- A busy router or an overloaded laptop.
The dish does not cancel those. Measuring “Starlink latency” on a bad Wi-Fi hop is measuring the hop.
This page does not rank Starlink against fiber, cable, or 5G. That comparison is Starlink vs fiber latency.
Why latency jumps
Jumpiness is the part people notice. Causes cluster.
Handover and a moving sky. The serving geometry changes. Brief delay spikes can appear even on a clear install. A spike you can repeat every few minutes in the app’s statistics is different from a one-off test blip.
Obstructions. A tree or eave that clips the view produces drops and delay spikes together. Confirm with the obstruction map.
Weather. Heavy rain and a snow-covered face degrade the radio. Overcast by itself is a weaker story—see clouds and rain.
Load. A saturated uplink (backup, cameras, a second call) queues packets. Queued packets are latency. This is why upload and latency get blamed for each other. If the queue is your backup, pause it.
Wi-Fi. Airtime contention and retries look like ping spikes. They vanish on Ethernet.
The other end. A distant VPN, a far game server, or a sick meeting region. Compare two destinations if you can: a nearby site versus the work VPN.
The device. Thermal throttling, battery saver, and a browser with a heavy meeting page add delay before the packet leaves.
Time of day. A busy cell can raise delay as well as cut throughput. Test the hour that failed.
If everything is slow, not just delay, start with why Starlink is slow and keep the order. Do not tune ping while the cable is loose.
Calls, games, and “feel”
Calls. You feel latency as stacked voices, awkward pauses, and lip-sync slip. You feel jitter as robotic audio. You feel loss as dropouts. A high download number will not save a call if delay is jumpy or upload is queued. Sit near the router, pause backups, and treat camera-plus-screen-share as two outbound jobs.
Games. Many titles tolerate delay that would annoy a competitive shooter. What hurts is a ping that leaps, plus loss. Whether Starlink is “good enough” depends on the genre and your tolerance; that decision belongs in Is Starlink good for gaming?, not in a latency definition page.
Browsing and video. Pages can feel instant with moderate delay if throughput is fine. Movies rarely care until delay turns into stalls from loss or a tiny pipe.
Feel is not a substitute for a sample. If the call was bad, measure at call time, on the call device’s network path.
How to measure latency correctly
Use TNET’s Starlink Speed Test for a consistent download, upload, and latency sample you can save. Then keep the method honest.
Wire when you can. Ethernet answers “what is the kit doing.” Wi-Fi answers “what is this room doing.” You need both only when the room is the complaint.
Do not ping the wrong thing. A game overlay, a VPN endpoint, and a speed-test server are different destinations. Compare like with like.
Repeat. One latency figure is a few seconds. Two or three runs at the problem hour beat one lucky millisecond count. If the number swings hard between back-to-back tests, believe the swing; that is jitter in disguise.
Unload the pipe. Pause backups. One device running the test.
Read jitter next to latency and check loss separately. A calm average with ugly jitter is not a calm path. See the results guide.
Change one variable. VPN off, then on. Next to the router, then the office. Clear weather versus the storm you blamed.
TNET will not invent a pass/fail millisecond cutoff. Fit is whether the job—call, game, remote desktop—stays usable on repeats. For expectation around throughput as well as delay, use Starlink internet speed. For regional context, use Starlink Coverage.
A dated weather-aware outlook for a place you will be is the Connection Outlook. It is a planning estimate, not a live ping.
A short latency checklist
Use this when a call or a game felt late and you want a cause, not a vibe.
- Write the time and the job that failed (call, game, remote desktop).
- Pause backups and extra cameras.
- Run the Starlink Speed Test on Ethernet or next to the router. Repeat once.
- Compare with a run in the problem room if the complaint is a room.
- Compare VPN on versus off if you work through a VPN.
- Check the Starlink app for obstruction or a weather-shaped outage while you test.
- If delay is calm on a wire and wild on Wi-Fi, stop at the room.
- If delay is jumpy on a wire and upload is busy, pause the backup and test again.
- If delay is jumpy on a wire with a clear dish and an idle house, review jitter, check the Starlink app for loss, then decide whether the far end (game region, meeting region) is the extra wait.
Two honest samples beat a night of overlays. If throughput is also down, leave this list and use why Starlink is slow.
What this page is not
This is not a ranked list of ISPs. It is not a claim that Starlink delay matches fiber. It is not a promise that games will feel local. Those buying questions sit on Starlink vs fiber latency and the gaming guide.
Latency on Starlink is delay on a LEO path plus every hop you added. It jumps when the sky, the queue, the room, or the far end jumps. Measure it on a wire when you can, at the hour you care about, more than once. Then fix the layer the split actually points to.
When you are ready to capture the number, run the Starlink Speed Test.