CS2 Input Lag: Reflex On or Off in 2026?
Input lag is the delay between moving your mouse and the pixel changing — lower means you peek first. Here is where it comes from, the genuinely contested Reflex vs -noreflex debate, and the latency setup for your rig.
Input lag is the gap between your mouse moving and the screen reacting. Shave it and you peek, flick and react a hair earlier than the enemy — over a match, that's rounds. Most of it comes from a few sources you can control. The one part that's genuinely contested in 2026 is NVIDIA Reflex, so let's settle what's known and what to test.
Where input lag comes from
- The render queue — frames buffered ahead of the GPU. The longer the queue, the more delay. This is what Reflex and Low Latency Mode target.
- Mouse polling rate — a 125 Hz rate adds noticeable latency versus 1000 Hz.
- Frame rate and cap — higher, stable frames lower latency up to a point; an unstable, uncapped rate can grow the queue.
- Sync mode — plain V-Sync adds the most latency; the G-Sync + V-Sync + Reflex + cap combo is the low-latency tear-free exception.
The Reflex debate
NVIDIA's official line is that in-game Reflex On delivers sub-15 ms PC latency on RTX 40-series and can cut system latency by up to 35%. But since early 2025, a vocal part of the competitive community calls Reflex "broken" in CS2 — reporting stutter — and runs -noreflex + Ultra Low Latency Mode in the driver + a driver FPS cap instead. Benchmarks point both ways depending on the rig, so this is genuinely rig-specific. Default to Reflex On; if you feel stutter, test the -noreflex setup and keep whichever is smoother for you.
The two setups, side by side
| Reflex On (default) | -noreflex camp | |
|---|---|---|
| In-game Reflex | Enabled | Disabled (via -noreflex) |
| Driver Low Latency Mode | Off (Reflex supersedes it) | Ultra |
| FPS cap | In-game, or refresh − 3 with G-Sync | Driver-level cap (NVIDIA App / RTSS) |
| Best for | Most players, RTX 40/50 | Players who feel Reflex stutter, RTX 30+ |
The free latency wins everyone should do
- 1000 Hz mouse polling (or higher on a strong CPU) and Enhance pointer precision off in Windows.
- Power Management Mode = Prefer Maximum Performance in the NVIDIA App so the GPU doesn't downclock.
- A stable FPS cap a little above your refresh — not so low your GPU sits idle. See the fps_max guide.
- One cap, one place — don't stack an in-game cap below your driver cap.
Bottom line
Do the free wins first — 1000 Hz polling, Prefer Maximum Performance, a stable cap, no acceleration. Then pick a lane: Reflex On for most, or -noreflex + Ultra Low Latency Mode if Reflex stutters on your rig. Let the Optimizer set it for your hardware, and if frames still feel rough, work through the stuttering fixes.
FAQ
Should I use Reflex On or -noreflex in CS2?
Default to in-game Reflex On — NVIDIA reports it cuts system latency substantially. But part of the community finds Reflex stutters in CS2 and prefers -noreflex plus Ultra Low Latency Mode in the driver. It is rig-specific, so test both and keep the smoother one.
Does -noreflex lower input lag in CS2?
It can on some rigs, when paired with Ultra Low Latency Mode in the driver and a driver-level FPS cap. On others, in-game Reflex On is lower latency. There is no universal winner — benchmark both on your own PC.
What mouse polling rate is best for CS2?
1000 Hz, or higher (2000–4000 Hz) only on a strong CPU. A 125 Hz polling rate adds noticeable latency compared to 1000 Hz, so never run below 1000 Hz for competitive play.
Does a higher FPS reduce input lag in CS2?
Up to a point — higher, stable frames shorten the render-to-display gap. But past your refresh the gains shrink, and an unstable uncapped rate can grow the queue. A stable cap plus Reflex or Ultra Low Latency Mode matters more.
What is the lowest-latency tear-free setup in CS2?
G-Sync On, V-Sync On in the driver, in-game Reflex On, and fps_max at your refresh minus 3. The cap keeps you inside the G-Sync range so V-Sync never engages, giving tear-free frames at near-minimum latency.