VPN speed: what slows your connection and how to measure it yourself

Lots of marketing numbers float around about VPN speed - often from one-off lab tests or theoretical simulations. Rather than give you numbers that won't match your connection, this guide explains what actually slows a VPN down and how to measure your own throughput loss with reliable tools. The goal: that you can read your own results, not memorize ours.

Why a VPN slows the connection

A VPN does two things that cost throughput and latency: it encrypts every packet (CPU work on both client and server) and it routes your traffic through an intermediate server before it reaches its destination. That detour adds physical distance, hence latency, and the encryption adds a small overhead to every packet.

It's unavoidable: there is no "lossless" VPN. But the size of that loss varies enormously with configuration. Well set up on a nearby server, a good VPN keeps most of your throughput; poorly configured or on a distant, overloaded server, the same app can halve your speed or worse. Understanding the levers below keeps you in the upper bracket.

The factors that really change your speed

The protocol (the #1 lever)

The VPN protocol selected in the app is the single most decisive factor. Modern protocols based on WireGuard - like NordVPN's NordLynx - are designed to be faster than OpenVPN, and that's a well-documented general fact:

• Modern cryptography: WireGuard uses ChaCha20, lighter on the CPU than OpenVPN's AES-256 on many devices.
• Minimal codebase: WireGuard is a few thousand lines where OpenVPN runs into hundreds of thousands, which reduces overhead.
• Kernel-space execution on Linux, faster than OpenVPN's userspace.

In practice, the typical ranking from fastest to slowest: NordLynx/WireGuard > OpenVPN UDP > IKEv2 > OpenVPN TCP. OpenVPN TCP is slowest because TCP adds its own reliability control on top of the tunnel; you only use it to traverse a corporate firewall that blocks UDP (TCP on port 443 then passes as web traffic). IKEv2 stays interesting on mobile because it handles Wi-Fi → 4G handovers well.

The setting to check in the NordVPN app: Settings → Connection → VPN protocol → explicitly force NordLynx. With the "Automatic" option, the app may fall back to OpenVPN on certain connections, costing throughput for nothing.

Distance to the server

The farther the VPN server, the higher the latency and the lower the throughput - that's physics. Light in fiber travels at roughly 200,000 km/s, so a transatlantic crossing (Paris → New York, ~5,800 km) imposes an incompressible round-trip of several tens of milliseconds, and a connection to Asia (Tokyo, Singapore, Sydney) adds even more. For maximum throughput, pick the nearest server that meets your need (unblocking a specific catalog, for example). For streaming, even a distant server is plenty: Netflix 4K needs only ~25 Mbps. Only competitive gaming truly suffers from the latency of very distant servers.

Server load and peak hours

A heavily used server shares its bandwidth across all active sessions. In the European evening (typically 9-11 PM), popular servers (US East Coast, London) and P2P servers receive far more connections, lowering per-session throughput. On top of that, your own ISP congests at the same hours, independent of the VPN. Two reflexes: switch server within the same region (often an immediate gain), and schedule large transfers outside peak hours when possible.

Encryption and filtering enabled

Filtering options like Threat Protection (ad/tracker/malware blocking) inspect traffic and add a slight overhead. It's an acceptable security/speed trade-off, but worth knowing if you're chasing every last Mbps. The encryption itself is non-negotiable: it's the whole point of a VPN.

Fiber, 4G, 5G: what to expect by network

Wired (fiber + Ethernet) is the most stable network and gives the best baseline. That's where a well-configured VPN keeps the most throughput, because the line suffers neither Wi-Fi nor mobile variability.

Wi-Fi introduces its own loss before the VPN even comes in: interference, distance to the router, channel sharing. If you measure a big drop "because of the VPN" over Wi-Fi, test wired first to separate the Wi-Fi share from the VPN share.

4G/5G is inherently more unstable than fiber: higher latency, jitter, bandwidth that varies with cell saturation. The relative loss from the VPN tends to be slightly larger there than on wired, which is normal - the tunnel amplifies existing instability. 5G offers a much better baseline than 4G where deployed, but coverage is uneven (on a high-speed train or in rural areas, throughput drops sharply, VPN or not).

How to measure your own VPN speed (reliable method)

The only numbers that matter are yours, on your line. Here's how to get an honest measurement in a few minutes.

1. Measure your baseline without the VPN. Disconnect the VPN and run speedtest.net (Ookla) or fast.com (Netflix). Note download, upload and ping. Wired if possible, for a clean reference.
2. Reconnect on a nearby server and run the exact same test, same tool, same test server.
3. Compare: the download/upload difference gives your throughput loss; the ping increase gives the latency added by the tunnel.
4. Run 2-3 successive measurements and keep the median - a single test is very noisy.
5. Repeat at different times (afternoon vs evening): network load changes the results a lot. To go further, iperf3 to a public server gives a more controlled measurement than web tests.

On NordVPN, you can also use our integrated speed test tool to compare with and without the VPN in seconds.

Diagnosis - why you measure slower than expected

If you measure a VPN markedly slower than your baseline, systematically check these six points in order.

Point #1 - Protocol. Are you on OpenVPN instead of NordLynx? Settings → Connection → VPN protocol → NordLynx. This is the #1 cause of an abnormally large loss.

Point #2 - Saturated server. Try another server in the same region. NordVPN has hundreds of servers per country; another one is often faster.

Point #3 - Wi-Fi vs wired. Part of the loss comes from home Wi-Fi, not the VPN. Test on wired Ethernet to compare your line's real baseline.

Point #4 - Distance. Connecting to Singapore from Paris? Normal to be slower. Physical distance is incompressible.

Point #5 - Peak hour. 8-11 PM in Europe = network congestion almost everywhere (at your ISP AND the VPN server). Retest at 2 PM to compare.

Point #6 - Threat Protection. In full mode, filtering adds slight overhead. Acceptable for security, but worth knowing if you optimize for speed.

If after all these adjustments you still see a very large loss on a local server, open a NordVPN support ticket - there may be a specific issue on the IP or server you're using.

What to remember

A VPN always slows things a little: that's the price of encryption and the detour through a remote server. But well configured, that cost is small and imperceptible for streaming, browsing and video calls. The levers that change everything: the protocol (force WireGuard/NordLynx over OpenVPN), distance to the server (closer is better), load (avoid saturated servers and peak hours), and Wi-Fi (wired gives the best baseline). For a detailed comparison between two services, our Surfshark vs NordVPN comparison lays the criteria side by side.

Most important: don't trust other people's numbers, measure your own. With a no-VPN baseline and then a test on a nearby server, you'll know in five minutes what your VPN really costs on your connection. Our integrated speed test tool makes that comparison easy.

Independent editorial assessment based on documented service capabilities, published independent audits and public benchmarks, as well as the general technical principles of VPN protocols. Commercial links carry the rel="sponsored nofollow" attribute; an affiliate commission may apply at no extra cost to the reader and with no influence on the rating.