A VPN dramatically slowing down internet speeds — pages crawling, video calls dropping, downloads taking forever — is the most common VPN complaint, and almost always fixable without giving up the privacy benefits. The big four causes: a server too far away, a bad protocol selection, the VPN provider being overloaded, or simply your local connection not being fast enough to handle encryption overhead. You’ll find the complete rundown in our Complete Guide to Fixing Windows, Browser, and Software Errors.
Quick reality check first: run a speed test without the VPN, then with. If the unencrypted speed is fine (say 50+ Mbps) and VPN brings it down to under 10 Mbps, there’s definitely something to fix. If your unencrypted connection is already 15 Mbps, expecting blazing speed through a VPN isn’t realistic — most home connections lose 15-30% of speed through any decent VPN, and that’s just the cost of encryption.
Try a closer server first
This is the single most impactful change and the one most people overlook. VPN speed is heavily affected by latency to the VPN server, and latency scales with distance. A server in your own country usually runs 2-3x faster than one across the world.
In your VPN client: change from “Auto-select” or “Best Server” to a specific server in your country or a neighbouring one. Run a speed test. If it’s noticeably faster — and it usually is — keep that server selected as default rather than letting the client auto-pick. Auto-selection algorithms in most clients prioritise load balancing over speed; manually picking the geographically closest healthy server is almost always better.
Special case: if you specifically need a VPN endpoint in a particular country (to access geo-restricted content), at least try multiple servers in that country. Server #3 in the list isn’t necessarily worse than server #1; some servers are heavily loaded while others nearby are idle.
Switch the VPN protocol
Most modern VPN clients support multiple protocols, and the default isn’t always the fastest. The current landscape:
| Protocol | Speed | Notes |
| WireGuard | Fastest | Modern, lightweight, lower CPU overhead |
| NordLynx / Lightway | Fast | Provider-specific WireGuard variants |
| IKEv2 | Fast on mobile | Great for switching networks; good speed |
| OpenVPN UDP | Moderate | Reliable, universal, but heavier |
| OpenVPN TCP | Slowest | Use only when UDP is blocked by firewall |
If you’re on OpenVPN TCP — switch to UDP or WireGuard immediately. The speed difference is often 2-4x. If your client doesn’t expose protocol selection, look in advanced settings or “Connection” settings. NordVPN, Surfshark, ExpressVPN, and ProtonVPN all let you change this.
WireGuard is the modern default for a reason — it’s significantly faster than older protocols while being equally secure. Use it unless you have a specific reason not to.
Server overload — the rush hour problem
Popular VPN servers get crowded at peak hours. Servers in the US and UK are often slow between 6-11pm local time because every user from those time zones is connecting after work. The same server at 3am might be 5x faster.
Most VPN clients now show “Server Load” as a percentage or coloured indicator. Pick a server under 40% load if possible. If your provider doesn’t show load: try several servers and use whichever is fastest at that moment.
For chronic congestion: consider whether your VPN provider just doesn’t have enough capacity. The cheaper VPNs often oversell their server capacity, and you end up sharing crowded endpoints with thousands of other users. Established premium providers (Mullvad, ProtonVPN, NordVPN) tend to maintain better server-to-user ratios.
Split tunnelling
Not everything you do needs to go through the VPN. Streaming Netflix, downloading large files, video calls — these are usually fine unencrypted (and the VPN often slows them down or breaks them). Split tunnelling lets you route specific apps or destinations through the VPN while everything else goes direct.
In your client: look for “Split Tunneling” or “App Routing” or “Bypass List.” Add high-bandwidth apps you don’t need encrypted (Netflix, Spotify, gaming clients, Zoom) to the bypass list. Your browsing stays VPN-protected; the bandwidth-heavy stuff doesn’t fight VPN overhead.
This essentially restores full speed for the use cases that suffer most from VPN slowdown without compromising the privacy aspects that matter to you.
Our guide on VPN connection problems covers the situations where the VPN won’t connect at all, which sometimes manifest first as severe slowness, and our DNS troubleshooting covers the DNS slowdown that often accompanies VPN use. For protocol details and security comparisons across the major VPN options, ProtonVPN’s documentation covers WireGuard versus OpenVPN tradeoffs in detail.
CPU bottleneck on older hardware
VPN encryption is CPU-intensive. On older machines (5+ years old, low-end laptops, mini PCs), the CPU itself becomes the bottleneck — you can’t go faster than the CPU can encrypt. Modern CPUs with AES-NI hardware acceleration handle VPN traffic with minimal overhead; older CPUs without it drop speed dramatically.
Check Task Manager → Performance → CPU while running a speed test through the VPN. If CPU usage spikes to 80-100% during the test, you’ve hit the CPU ceiling. Possible fixes:
- Use WireGuard (significantly less CPU-intensive than OpenVPN)
- Run the VPN on the router instead of individual devices — offloads encryption from the laptop to dedicated router hardware
- Upgrade the hardware eventually if the machine is otherwise slow too
For a 4-5 year old laptop, this is rarely the limiting factor with WireGuard. With OpenVPN on older hardware, yes, very plausible.
MTU and packet size tuning
This is more technical but worth knowing. VPN tunnels add encryption headers that take up space inside each network packet. If the resulting packet exceeds the maximum MTU (Maximum Transmission Unit) for any hop along the way, it gets fragmented — which kills speed.
Default MTU on most networks is 1500. Inside a VPN tunnel, useful payload drops to around 1380-1450 depending on protocol. If you can configure MTU in your VPN client, setting it to 1400 (or 1380 for OpenVPN) often improves throughput, especially on connections that have lower MTUs than expected (some mobile data networks, some PPPoE-based broadband).
To find your effective MTU: Command Prompt → ping -f -l 1472 8.8.8.8 → if it says “Packet needs to be fragmented but DF set,” lower the number by 10 and try again, until it succeeds. Add 28 to that number to get your effective MTU. Then set your VPN’s MTU to about 50 below that.
This is overkill for most users but worth knowing if you’re stuck at low speeds despite all other optimisations.
Local network and ISP interference
Sometimes the bottleneck isn’t your VPN at all — it’s your local network or ISP. Common scenarios:
Wi-Fi congestion: 2.4 GHz networks in crowded neighbourhoods (lots of neighbouring routers, microwaves, Bluetooth devices) lose substantial speed during peak hours. Switching to 5 GHz, or hardwiring with Ethernet for a test, isolates whether Wi-Fi is the problem.
ISP throttling: Some ISPs deprioritise VPN traffic, especially during peak congestion. Symptoms: VPN speed drops at specific times of day correlated with overall internet usage. Hard to prove, harder to fix — switching providers or using a different VPN port (some VPNs let you tunnel over port 443 to disguise traffic as HTTPS) sometimes helps.
Router can’t handle VPN throughput: if you’re running the VPN on the router itself, cheap routers have weak CPUs that bottleneck at 30-50 Mbps even when your line is 500+ Mbps. Higher-end routers (or dedicated VPN-capable router models) handle 200+ Mbps. ASUS RT-AX86U, GL.iNet Flint, and Synology RT2600ac are popular options for router-level VPN.
Quick benchmarks to know if your speed is reasonable
| Native speed | Reasonable VPN speed | Probably bad if under |
| 1 Gbps | 500-800 Mbps (WireGuard) | 200 Mbps |
| 500 Mbps | 300-450 Mbps | 150 Mbps |
| 100 Mbps | 70-90 Mbps | 40 Mbps |
| 50 Mbps | 35-45 Mbps | 20 Mbps |
| 25 Mbps | 18-22 Mbps | 10 Mbps |
These are rough expectations for WireGuard on modern hardware. OpenVPN typically delivers about 60-70% of WireGuard speeds. If your VPN speed falls well below these benchmarks, something is correctable.
When to use a different VPN provider
If you’ve tried different protocols, different servers, split tunnelling, and the speed is still bad: it might just be the provider. Free VPNs are notorious for being slow — they have to be, since they fund themselves through limited bandwidth and ads. Some budget VPNs (£2-£3/month tier) cut corners on server capacity.
Premium VPNs with consistently good speed reputations in 2026: Mullvad (£5/month, no-frills but very reliable), ProtonVPN (free tier exists, paid plans fast), NordVPN (heavy advertising but the speeds back it up), Surfshark (good value, fast network). If you’re paying for a premium service and seeing speeds 3-4x slower than these benchmarks, switching providers usually solves the problem more reliably than further configuration.
Conversely, if you’re using a free VPN and complaining about speed: that’s the bargain you signed up for. Move to a paid tier or accept the limitation.
Practical real-world scenarios
A few patterns worth knowing because they come up often:
“VPN is slow on hotel/airport WiFi”: public WiFi networks frequently throttle, deep-packet-inspect, or block VPN traffic. There’s often nothing wrong with your VPN itself; the network is fighting it. Switch to mobile data for the VPN traffic if speed matters.
“VPN is fast on speed test but slow on real sites”: speed tests use specific server endpoints close to major data centres. Real sites use CDNs that pick endpoints based on your apparent location (which is the VPN endpoint, not you). A US-based CDN serving content via UK VPN might be far slower than the same content delivered to your real UK home connection.
“VPN was fast yesterday, slow today”: server load fluctuates, ISP congestion shifts, time of day matters enormously. Switch to a different server before assuming anything’s broken.
“VPN is slow only on specific websites”: some sites actively detect and throttle VPN traffic (streaming services especially), or apply rate limits when they detect data centre IPs. Not a VPN configuration issue — a policy at the destination site.
The honest summary
The four highest-impact changes you can make: switch to WireGuard, pick a closer server, enable split tunnelling for high-bandwidth apps you don’t need protected, and check whether your client lets you choose between server-load levels. Combined, these typically restore VPN speeds to 70-90% of your native connection on modern hardware.
If you do all of that and speed is still bad, the constraints are real and structural: ageing CPU, ISP throttling, or just a connection too slow to handle encryption gracefully. At that point, the realistic option is either to accept the loss as the price of privacy, or invest in better hardware/connection. There’s no magic configuration that overcomes physical limits.
Mobile and laptop differences
Laptops and desktops on home WiFi or Ethernet handle VPN traffic differently from phones and tablets on mobile data. Mobile carriers often apply additional traffic shaping that affects VPN performance specifically — speeds that look fine on home WiFi might be terrible on 4G/5G.
For mobile users: IKEv2 is usually the best protocol because it handles network switching (WiFi to cellular to a different cellular cell) gracefully. WireGuard works on mobile but can disconnect more often when networks change. Battery usage also matters — WireGuard is the lowest-impact on mobile battery; OpenVPN is the heaviest.
If a VPN absolutely destroys your mobile data speeds (going from 100 Mbps unencrypted to 5 Mbps over VPN), check whether your carrier is identifying VPN traffic and throttling it. Some carriers do this on cheaper plans. Switching ports (some clients let you tunnel over HTTPS port 443) can sometimes evade this.
Final reality check
VPNs were never going to be free — they add encryption overhead, route traffic through extra hops, and depend on the provider maintaining good infrastructure. Reasonable expectations: 70-90% of your native speed on modern hardware with WireGuard, somewhat less on older hardware or with OpenVPN. If you’re hitting those numbers, the VPN is doing its job correctly.
Below those numbers, the fixes above usually help. Above those numbers — congratulations, you have an unusually well-optimised setup. Either way, the choice between speed and privacy isn’t binary; tuning the configuration with the practical priorities above gets you most of both.
One small but effective tip for people who use VPNs heavily: keep a “favourites” list in your client of 4-5 servers you know work well, in different countries. When your default server gets slow, switching to a known-good alternative takes 5 seconds instead of cycling through random options. Most premium VPN clients support this; if yours doesn’t, just note them in a text file. Over time you learn which specific servers in which cities consistently deliver good speed at your typical use hours, and those become your go-to choices. You might also run into VPN Slowing Internet Speed.







