How to Reduce Input Lag: The Complete Latency Guide

By Joy Jacob · Updated 2026-06-28 · 4 min read

Input lag is the delay between your hand moving and the screen reacting — the gap between a click and the muzzle flash. It's invisible until you feel it, and then it's the difference between winning a duel and swearing at your monitor. The frustrating part is that it isn't one number you can turn down; it's a chain of small delays, each adding a few milliseconds, stacked from the moment you move the mouse to the instant a photon hits your eye. Fix the wrong link and you get nothing. This guide walks the whole chain in milliseconds, shows you where the real time is hiding, and lists the changes that actually move the needle — then hands you a calculator to budget your own latency.

The input-to-photon chain

Here's the journey of a single mouse movement, stage by stage, with rough time contributions on a decent gaming setup:

Stage	What happens	Typical ms
Peripheral polling	Mouse reports its position; you wait up to one polling interval	1–8 ms
USB & OS	Signal travels the bus and through the input stack	~1 ms
Game engine	Input is sampled, game state simulated	1–8 ms
Render queue	Frames buffered ahead of the GPU (the big hidden one)	0–40 ms
GPU render	The frame is actually drawn	3–16 ms
Display scanout	Frame is sent to the panel; ~half a frame on average	2–8 ms
Pixel response	The liquid crystal physically changes color	1–10 ms

Notice which stage has the widest range: the render queue. It's where a badly configured system hemorrhages latency, and it's the one most people have never heard of.

The render queue, explained

To keep the GPU fed and the frame rate smooth, the CPU can queue up several frames' worth of work in advance. That buffering is great for average FPS but terrible for latency: an input you make now might not be reflected until the GPU chews through the frames already waiting in line. If three frames are queued at 60 FPS, that's ~50 ms of pure delay before your action even reaches the screen.

The fix is to shorten or bypass that queue. Low-latency modes do exactly this — they cap the number of pre-rendered frames and dynamically pace the CPU so it submits work "just in time" instead of racing ahead.

The biggest wins, in order

1. Cap your FPS below your max with a limiter. Keeps the GPU under 100%, collapses the render queue. The most effective single change for most players.
2. Enable a low-latency / Reflex mode. NVIDIA Reflex, AMD Anti-Lag, or the driver's "Ultra" low-latency setting pace frame submission to slash queue delay. If a game has Reflex, turn it on — it's close to free latency.
3. Kill VSync, or pair VRR with low-latency. Traditional VSync adds a frame or more of buffering. Best setup: a variable refresh display (G-SYNC/FreeSync) + in-game FPS cap a few frames below refresh + low-latency mode + VSync off.
4. Use a high-refresh, low-response display. A 240 Hz panel scans out a frame in ~4 ms versus ~16 ms at 60 Hz. Fast IPS/TN pixel response beats slow VA for transitions.
5. Go wired and high-polling. A 1000 Hz mouse adds at most 1 ms of polling delay; a 125 Hz one adds up to 8 ms. Wired removes wireless negotiation overhead (good 2.4 GHz wireless is close, but wired is a safe default).
6. Raise your frame rate generally. Higher FPS means each frame is fresher; latency scales roughly with frame time even before the queue.

Refresh rate versus response time

People conflate these. Refresh rate (Hz) is how often the panel draws a new image — it sets the scanout portion of your budget and the freshness of what you see. Response time (ms) is how fast an individual pixel physically changes shade — it causes smearing and ghosting, and adds its own tail of delay. You want both good: high refresh for fresh frames, low response for clean transitions. A deeper walk-through lives in our refresh rate guide.

What doesn't help (much)

• Buying a faster mouse when yours already polls at 1000 Hz. You've maxed that stage; the ms is elsewhere.
• Cranking DPI. Sensitivity is not latency — see DPI and sensitivity explained. Higher DPI moves the cursor further, it doesn't move it sooner.
• Overclocking for the sake of it if your GPU is already under 100% at a capped frame rate; the queue, not raw horsepower, is your bottleneck.

Budget your latency

The calculator below sums the main stages of your chain into an estimated total, with a per-stage breakdown so you can see where your milliseconds are going. Pick your refresh rate, panel response, and mouse polling, and toggle whether you run a frame limiter. Numbers are reasonable approximations, not lab measurements.