VPN Latency Explained: Why Ping Changes

You open a game, a video call or a remote desktop session, switch the VPN on, and the numbers change. The ping in the corner jumps from 25 to 70. Pages feel a fraction slower to start. Nothing is broken, yet something has clearly changed. That something is latency, and it behaves very differently from the "speed" number most people watch.

This article explains latency in the context of a VPN. It covers what latency is made of, why a tunnel adds to it, how jitter and packet loss relate, what typical numbers mean for different activities, and what you can and cannot do about it. It deliberately avoids repeating the broader picture in why internet speed changes with a VPN and the activity-by-activity view in does a VPN reduce ping. Here we go one level down, into the milliseconds themselves.

Latency in one picture: the round trip

Imagine you send a postcard and wait for a reply postcard. The time between posting yours and receiving theirs is the round trip. Latency, as the term is used on networks, is that round trip for a tiny piece of data. Cloudflare's definition of latency is a good neutral reference if you want another explanation.

Two points are worth fixing in your mind:

• Latency is a time, not a rate. It is measured in milliseconds (ms). Lower is better.
• Ping is a measurement of latency. A ping tool sends a very small message to a destination and times how long the reply takes. The number you see is a round trip, not a one-way trip.

Bandwidth, the "Mbps" number, answers a different question: how wide is the pipe. You can have a very wide pipe that is very long. Water still takes time to travel along it. A large download from a nearby server and a tiny request to a far server can behave in opposite ways because they depend on different things.

Where every millisecond comes from

A round trip is a sum of delays. Understanding them makes the VPN effect easy to predict.

Source of delay	What it is	Does a VPN change it?
Propagation	Signals take time to cross distance; light in fibre covers roughly 200 km per millisecond	Yes, if the detour is longer
Transmission	Time to put the bits on the wire; small for small packets on fast links	Slightly
Processing	Routers and devices handle each packet	Yes, adds encryption work
Queuing	Waiting in line when a link is busy	Depends on the server and your line
Last mile	Wi-Fi, mobile radio or the cable to your house	No, same as before

Propagation is the part you cannot argue with. Distance costs time. A trip to a server on the other side of the world can cost well over a hundred milliseconds even on a perfect network, simply because of geography and the winding path cables take. Queuing is the part that varies most: a busy link is like a queue at a checkout, and the wait changes minute by minute.

What a VPN adds to the round trip

Without a VPN, the round trip runs from your device to the destination and back. With a VPN it runs from your device to the VPN server, from the VPN server to the destination, and then all the way back along the same path. In effect, you replace one leg with two.

A simple worked example, using illustrative numbers rather than measurements:

• Direct to a game server: 30 ms.
• Your device to the VPN server: 15 ms.
• VPN server to the game server: 25 ms.
• Tunnel processing at both ends: around 2 to 5 ms.

With the VPN on, the total is roughly 15 + 25 + processing, which comes to about 42 to 45 ms, compared with 30 ms without. The extra 12 to 15 ms exists because the VPN server is not on the straight line between you and the game server. If the VPN server sat on that straight line, the extra would be mostly processing, a few milliseconds. If it sat far away, the sum could be several times larger.

This is why the rule of thumb for lowering VPN latency is geometry: pick a location that keeps the total path short, which is usually near you or near the service you use. How to choose a VPN server location develops that idea. It is also why latency can occasionally improve: if your direct path to a destination is unusually roundabout, a VPN server with a good onward connection can sometimes shorten it.

For a refresher on what the tunnel itself is, see what is a VPN tunnel. The connection-set-up delay at the start of a session is a one-time cost and is separate from steady round-trip latency.

Jitter and packet loss: the other two numbers

Latency has two companions that decide how a connection really feels.

Jitter is the variation in latency. Suppose five consecutive pings read 40, 41, 40, 42, 40 ms. That is a stable line with low jitter. Now suppose they read 25, 90, 30, 140, 35 ms. The average might be similar, but the second line will feel terrible in a call, because packets arrive in uneven bursts. Real-time apps use small buffers to smooth this, but the smoothing adds delay of its own.

Packet loss is the share of packets that never arrive. Even one or two percent can cause audible glitches in a call and rubber-banding in a game, because the missing data has to be resent or guessed.

A VPN can influence both. A congested VPN server or a weak Wi-Fi link tends to raise jitter. Some tunnels that run over reliable, ordered connections can turn packet loss into extra delay, because a lost packet holds up everything behind it until it is resent. Newer transport designs, such as QUIC described in RFC 9000, are built partly to reduce that kind of stall, though what a given VPN app uses is shown in the app, not something to assume.

Reading ping numbers in plain language

Treat the following as rough orientation, not rules. They depend on the application and on how forgiving it is.

Ping range	Typical feel
Under 30 ms	Excellent; almost instant for any use
30 to 60 ms	Very good; fine for competitive play and calls
60 to 100 ms	Good for browsing, calls and most games
100 to 150 ms	Noticeable in fast games and remote desktop
150 to 250 ms	Calls get awkward, games feel laggy
Over 250 ms	Frustrating for anything interactive

Notice how little all this matters for normal browsing. Opening a web page involves several round trips, so it does feel a hundred-millisecond difference, but you rarely notice a difference of 20 ms. The same 20 ms in a fast shooter can be the difference between a hit and a miss.

Also remember what you are measuring. A ping to a nearby test address says little about latency to the particular service you use. When possible, measure to the destination that matters, such as the game's region or your company's service.

Three scenarios

A call that suddenly feels awkward. Lina notices that people keep talking over each other on her calls once she switches the VPN on. Her ping went from 35 to 110 ms with a lot of variation. She changes from a faraway location to a closer one and the ping settles near 55 ms. The conversation flows again. The cause was distance, not bandwidth.

A developer on a remote desktop. Marco connects to an office machine. Every typed character appears a beat late. His round trip adds up: home Wi-Fi, then the VPN server, then the office network. He moves to a cable connection and the Wi-Fi leg shrinks by 15 ms. The VPN part stays the same, but the total is more comfortable. Fixing the leg you control is often easier than fixing the VPN.

A family movie night. The family streams a film over the VPN. Latency rises by a visible amount, but the video starts after a moment and plays smoothly. No one notices, because the player buffers ahead and the stream cares about throughput, not round-trip time. Latency matters much less here than for a call.

How to measure latency fairly

You do not need special software. Most operating systems include a ping command, and many apps display ping directly.

1. Pick the destination that matters to you.
2. Measure with the VPN off, 10 to 20 pings, and note the typical value and the range.
3. Connect the VPN, wait until it is steady, then measure to the same destination.
4. Compare both the average and the spread. If the average rose by 15 ms but the spread doubled, the jitter is the bigger issue.
5. Repeat at another time of day to see whether the change is stable.

Be careful with one thing: some networks treat ping traffic differently from other traffic, so a ping tool is an indicator rather than a perfect copy of how your real application behaves. If possible, also look at the latency your own application reports. For general advice on checking whether a tunnel is working as expected, see why is my VPN so slow.

What helps, and what does not

Things that often help:

• A location closer to you or to the service. Geometry is the biggest lever.
• A cable or a better Wi-Fi spot. This removes jitter from the last mile.
• Quiet background traffic. Large uploads and updates fill queues on your own line and add delay to everything else, a problem sometimes called bufferbloat.
• Off-peak hours. Queues are shorter late at night.
• Leaving settings at their defaults, unless support suggests a change. If you do change a setting, change one at a time.

Things that rarely help:

• Paying for more bandwidth when the problem is latency. A bigger pipe does not shorten the trip.
• Restarting everything repeatedly. One restart is reasonable; ten are not a fix.
• Chasing the lowest possible number. Below the level your application needs, there is no benefit.

Limits to be honest about

Latency cannot be reduced below the physical limits of distance. A VPN adds at least a small amount of processing and at least some detour. Different networks, devices and times give different results. Some apps are more tolerant than others, and a few are sensitive enough that you may prefer to leave the tunnel off for them while keeping it on for everything else, using split tunneling if your app offers it. A VPN is a protection tool first; if the latency cost is too high for a particular activity, it is reasonable to decide that the activity does not need it.

Summary

Latency is the round-trip time of a small piece of data, measured in milliseconds. A VPN replaces one leg of the journey with two and adds a little processing, so latency usually rises somewhat, mostly because of geometry. Jitter and packet loss decide how smooth the connection feels. Pick a sensible location, fix the last mile, and judge the numbers against what your activity needs.

If you want to see how your own ping behaves, the pricing page lists short prepaid passes that are enough for a few days of testing.