Your ping is set by the distance to the server, and no software fixes that. Everything else is in your hands: use a cable instead of Wi-Fi, fix your router’s bufferbloat, and pick the nearest server. A steady 60 ms plays better than a 30 ms that jumps around.

Three different numbers we all call “lag”

When someone says they have lag, they could mean three separate things, each with its own fix.

• Ping (latency): how long a packet takes to reach the server and come back. It’s the number in the corner of your screen.
• Jitter: how much that ping varies from one packet to the next. A ping of 40 ms that stays at 40 is comfortable; one swinging between 25 and 120 is what kills you behind cover.
• Packet loss: packets that never arrive. You see it as teleporting enemies and shots that don’t register.

Jitter and loss bother you more than average ping, and they’re exactly what a normal speed test won’t show you. That’s why plenty of people have “gigabit fibre” and still play badly.

There’s a fourth number that isn’t network at all: system latency, from the moment you move the mouse to the moment the monitor draws it. NVIDIA separates it from ping explicitly in its Reflex documentation: Reflex shortens the GPU’s render queue and does nothing to the network.

What you can’t change: physics

Light in fibre travels at roughly 200,000 km/s — slower than in a vacuum, because of the refractive index of glass — or about 200 km per millisecond. Since ping is a round trip, every 100 km of distance costs 1 ms as an absolute floor, before a single router enters the picture.

London to New York is about 5,570 km: physics alone imposes roughly 56 ms against a server there, and the real route is never a straight line. London to Frankfurt is about 640 km, so around 6 ms of floor.

So before you touch anything: check where the server you’re playing on actually is. Switching region in the game menu does more than every optimisation combined, and it’s the first thing almost nobody checks.

Cable or Wi-Fi: what really changes

Today’s Wi-Fi is nothing like it was ten years ago, but it still loses on the thing that matters here. It’s a shared, half-duplex medium: only one device transmits on that channel at a time, and everything else — yours and your neighbours’ — waits its turn. That wait is variable, and variable means jitter. A cable has its own lane, in both directions, all the time.

In practice:

• Ethernet cable: the lowest ping and, above all, the flattest. Cheap Cat5e is plenty; you don’t need Cat8.
• Wi-Fi on 5 or 6 GHz, close to the router: perfectly playable. Adds a few milliseconds and some jitter.
• Wi-Fi on 2.4 GHz: the most crowded channel in the house, shared with the microwave and half the block. Worst case.
• A repeater or a wirelessly backhauled mesh node: every wireless hop adds waiting again, which is why Rtings advises against gaming through one.
• Powerline adapters: fine when there’s no alternative, but in PC Gamer’s testing they produced the worst latency spikes. They depend on your home’s electrical wiring, so they’re unpredictable.

Wi-Fi 7 with MLO genuinely improves this: field trials by the Wireless Broadband Alliance with CableLabs and Intel measured 35% to 48% lower application latency and up to 40% less jitter. But it needs a compatible router and adapter, and even then it still trails a five-dollar cable.

The culprit almost nobody measures: bufferbloat

This is the “my ping is fine until someone starts streaming” case.

When your connection saturates, the router piles packets into a queue instead of dropping them. Your game packets — tiny and in a hurry — end up queued behind a flood of video. The result: a 25 ms ping at idle that spikes to 200 or 400 ms the moment someone uploads a file.

You can measure it free with Waveform’s bufferbloat test or the loaded latency section of Cloudflare’s speed test. What you want is the gap between idle latency and latency under load: if it climbs more than 30-40 ms, you have bufferbloat.

The fix isn’t buying more bandwidth. It’s enabling SQM (Smart Queue, adaptive QoS, or by its algorithm, fq_codel or CAKE) and setting the upload and download caps at 85-95% of your measured real speed. That trim is the whole trick: the queue stays in your router, which knows how to manage it, instead of in the ISP’s box, which doesn’t.

Two warnings. On a low-end router, turning on SQM can cut your peak speed — managing the queue costs CPU. And if your ISP’s router doesn’t offer the option, the only real way out is putting your own behind it.

What to do, by symptom

What you’re seeing	Most likely cause	What to do
High ping all the time, but steady	Distance to the server	Change region in the game. There’s no technical fix
Fine until someone else uses the internet	Bufferbloat	Enable SQM/QoS and cap at 85-95%
Random spikes, fine the rest of the time	Wi-Fi interference	Move to cable, or to 5/6 GHz near the router
Teleporting and shots that don’t register	Packet loss	Check cable and connectors; if it persists, it’s your ISP
Feels heavy even though ping is low	System latency, not network	Monitor Hz, V-Sync off, Reflex/Anti-Lag
Only happens in one game	The game’s servers or netcode	Check the service status before blaming your network

As a rough guide: under 50 ms plays anything, 50 to 100 ms is fine for everything except competitive shooters, and packet loss should be 0%. Sources don’t agree on an exact jitter number, so use the test that actually works: compare your idle ping with your in-game ping. If it doesn’t move, you’re fine.

What doesn’t work (or is just marketing)

In order of money wasted for nothing:

• Buying more bandwidth. An online game uses almost no data: going from 300 Mb to 1 Gb won’t drop your ping by a single millisecond. It only helps if you were saturating, and SQM is cheaper for that.
• Your router’s “gaming mode”. Rtings tested 11 routers and found no meaningful difference in pure latency between modern models. The only thing that adds value is QoS/SQM, and ordinary routers without the label have it too.
• Cat7 or Cat8 cables. Category affects bandwidth, not latency. Cat5e in good condition performs the same.
• “Ping boosters” and gaming VPNs. They add a detour. They only help if your ISP’s default route to that server is worse than theirs; if your route is already direct, the detour can only make it worse.
• Windows “optimisers”. Editing the registry or disabling Nagle’s algorithm changes nothing measurable on a healthy connection.

What is free and usually does show up: close whatever is syncing in the background before you start playing, and reboot the router if it’s been on for months.

Frequently asked questions

What counts as a good ping for gaming? Under 50 ms is comfortable in any game; 50 to 100 ms is playable except in competitive shooters. Stability matters more: a flat 60 ms plays better than a 30 ms that spikes to 150.

Does an Ethernet cable lower ping compared to Wi-Fi? It lowers average ping a little, but its real advantage is jitter: on Wi-Fi your packet waits its turn and that wait varies, while a cable always has its own lane. You notice it most in a house full of connected devices.

Why do I have gigabit fibre and still lag? Because bandwidth and latency are different things — an online game uses very little data. The usual culprit is bufferbloat: ping is fine at idle and spikes when the line gets busy.

Are ping boosters or gaming VPNs worth anything? Only if your ISP’s default route to that server is worse than the service’s. They don’t shorten physical distance, and if your route is already good, the detour makes it worse. Measure it during the free trial before paying.

Does NVIDIA Reflex lower ping? No. Reflex reduces system latency — mouse to screen — by shortening the GPU’s render queue, and NVIDIA explicitly separates that from network latency.

What to look at next

If measuring shows the problem wasn’t your network, the screen is next: 144 Hz or 240 Hz: what you actually notice has the maths on how much each refresh jump shaves off, and how to monitor FPS and temperatures for free gives you the tools to tell whether the GPU, the CPU or heat is holding you back. If the game feels heavy with good ping and good FPS, check frame generation in DLSS, FSR and XeSS explained.

And if you’re building or upgrading, the Ethernet port and the wireless adapter are decided when you pick which motherboard to buy. On laptops, our gaming laptop guide covers which models keep a network port and which force you onto a USB adapter.

Sources consulted

• Rtings – Buying A Gaming Wi-Fi Router Won’t Help Your Ping: Understanding The Limits Of Wi-Fi For Gaming
• NVIDIA – Reflex Low Latency Platform (official documentation on system latency vs network latency)
• PC Gamer – Here’s how playing on Wi-Fi hurts your game
• Wireless Broadband Alliance – WBA Validates Wi-Fi 7 MLO Reliability with CableLabs and Intel
• Waveform – Bufferbloat and Internet Speed Test
• Cloudflare – Speed Test (latency under load)
• Bufferbloat.net – The CAKE and fq_codel queue management algorithms

Published 12 November 2026. The distance-based latency figures are calculated from the propagation speed in fibre (~200,000 km/s) and straight-line distances between cities.