Mouse acceleration is one of the oldest and most heated debates in FPS - and one of the most thoroughly answered by data, biomechanics, and pro player consensus. Yet every year the question resurfaces on Reddit, YouTube, and Discord because a handful of legendary players (notably some Quake veterans) defend it. This guide walks through the science, the technical layers of acceleration, the empirical pro distribution, and the verdict for CS2, Valorant, Apex, and Overwatch players in 2026.
Mouse acceleration is a non-linear input curve where the relationship between physical mouse movement and on-screen cursor displacement depends on speed. With acceleration enabled, a 10 cm slow swing produces less cursor displacement than a 10 cm fast swing.
The opposite is linear mapping (also called "no accel", "1:1 input", "raw input"): the cursor moves a fixed distance per physical mouse movement, regardless of swing speed.
| Layer | Default State | Recommended for FPS |
|---|---|---|
| Windows Enhanced Pointer Precision | ON by default | OFF |
| Mouse driver software (Logitech G Hub, Razer Synapse) | OFF by default | OFF |
| In-game acceleration (m_customaccel, mouse_acceleration) | OFF by default in modern games | OFF |
Every modern FPS supports raw input (bypassing OS-level acceleration). But Windows EPP is enabled by default on a fresh install and many players never disable it.
Muscle memory in FPS works by encoding a stable mapping: hand movement X always produces in-game movement Y. This relationship must be consistent across hours of repetition for the cerebellum to compile it into procedural memory. Acceleration breaks this:
The brain cannot form a single learned model. It has to learn the displacement-vs-speed curve, which is far harder and never reaches the same precision ceiling. After 100 hours, a linear player has 100 hours of pure flick memory. An accel player has 100 hours of context-dependent flick memory that still requires conscious correction.
From publicly available pro settings databases (prosettings.net, prosettings.video):
| Game | Total Pros Surveyed | Use Acceleration | Linear |
|---|---|---|---|
| CS2 | 247 | 3 (1.2%) | 244 (98.8%) |
| Valorant | 189 | 0 (0%) | 189 (100%) |
| Apex Legends | 156 | 2 (1.3%) | 154 (98.7%) |
| Overwatch 2 | 124 | 0 (0%) | 124 (100%) |
| COD MW3 / Warzone | 198 | 0 (0%) | 198 (100%) |
The pattern is overwhelming. The handful of acceleration users (such as some old-school Quake-rooted players) learned on it 15-20 years ago when displays and DPI options were limited - their muscle memory cannot be retrained without massive performance regression.
Some players genuinely prefer the feel of acceleration. Common reasons:
The first two are solved by buying a larger mousepad (XL or XXL, ~20-25 USD) and using a slightly higher in-game sensitivity. The third and fourth are inertia, not preference - players who switched to linear after 6-12 months of relearning almost universally report better consistency.
For casual FPS (Krunker.io, Battlefield campaign, etc.), acceleration is harmless. The skill ceiling is lower; muscle memory has less compound effect. If you only play casually and like the feel, it's fine. But if you intend to climb ranked in CS2, Valorant, Apex, or any modern competitive FPS - disable it.
Both Aim Lab and Kovaak's default to no acceleration. Confirm in your settings:
If you train with no acceleration but play with it, transfer is broken. Match them. Always.
Raw Input is a Windows API call that bypasses OS-level mouse processing - including acceleration. When a game uses raw input (`m_rawinput 1` in CS2, default in Valorant/Apex), it receives mouse data directly from the device, ignoring Windows mouse pointer settings.
Raw input is on by default in every modern competitive FPS. You should not need to enable it manually except in CS2 where the convar exists as a safety net.
| Term | Meaning | Where Set |
|---|---|---|
| DPI | Mouse sensor resolution (dots per inch of physical movement) | Mouse driver / on-mouse button |
| In-game sensitivity | Software multiplier on raw mouse movement | Game settings menu |
| eDPI | DPI × in-game sensitivity (effective sensitivity) | Calculated, not set |
| Acceleration | Non-linear scaling based on movement speed | Windows / driver / in-game (all should be OFF) |
| Polling rate | How often the mouse reports position to PC (Hz) | Mouse driver / firmware |
For competitive FPS, 400-1600 DPI is the pro range. Lower DPI (400-800) gives slightly smoother sensor tracking on older sensors; modern sensors (PAW3950, PAW3395, Hero 2) are flawless to 8000+ DPI. Choose any DPI in 400-3200 and adjust in-game sens to taste.
| Pro | DPI | In-Game Sens | eDPI |
|---|---|---|---|
| s1mple (CS2) | 400 | 3.09 | 1236 |
| donk (CS2) | 400 | 2.0 | 800 |
| TenZ (Valorant) | 800 | 0.4 | 320 |
| ImperialHal (Apex) | 800 | 1.4 | 1120 |
If you have been playing with acceleration and want to switch:
The transition is painful but pays back permanently.
On controller-based and mobile FPS, acceleration is built into aim-assist curves and cannot be disabled. The distinction is moot. This guide applies to PC mouse-and-keyboard play only.
Mouse acceleration in 2026 competitive FPS is a settled question. Disable it at every layer. The 1-2 percent of pros who use it are exceptional cases with two decades of accel-trained muscle memory. For a new or improving player, linear input is the unambiguous correct choice. Spend the disable-acceleration five minutes, never think about it again, and benefit for the rest of your FPS career.
Mouse acceleration in Windows was designed for productivity, not gaming. The Enhanced Pointer Precision algorithm assumes the user is performing pointing tasks where a small precise click is needed (clicking a desktop icon) but also wants fast cursor traversal across a large screen. Acceleration solves this by giving precision at slow speeds and large coverage at high speeds.
The problem: this assumption breaks down completely in FPS where the user wants consistent mapping between physical hand movement and virtual aim displacement. The default that helps office work hurts gaming. Microsoft has never removed it (backward compatibility) but every modern gaming-focused setup disables it.
A small group of professional players use acceleration intentionally. Most are Quake-rooted players from the 1999-2010 era when:
Players who learned in this environment developed acceleration-encoded muscle memory. Re-training them to linear would cost months of regression for a player whose career is built on accel-trained habits. Some elected to keep it; their performance is despite the choice, not because of it.
Some gaming mouse driver suites include optional acceleration features distinct from Windows EPP. Logitech G Hub has "Pointer Behaviour" settings that include subtle acceleration curves. Razer Synapse offers "Acceleration" sliders. Most defaults are off, but verify:
If your driver suite recently updated, double-check these settings - some updates reset preferences to factory defaults.
Counter-Strike 1.6 had a known acceleration bug where m_filter and m_customaccel interacted unexpectedly, producing inconsistent input even when settings were "off". Source engine fixed this; CS:GO and CS2 use raw input by default with no engine-level acceleration unless explicitly enabled via console.
To verify your CS2 configuration is clean: open console (`
Open console (~ key by default, or backquote, depending on keyboard layout) and check:
If any of these are wrong, set them via console and add to autoexec.cfg.
Apex has a unique "Response Curve" setting that controls how stick input maps to aim - relevant primarily for controller players but also exposed for mouse. For mouse input, set to "Classic" (also called "Linear" in some versions). The other curves (Quadratic, Cubic, Custom Classic) introduce non-linearity that approximates acceleration.
Note: aim assist on Apex (for controller) inherently uses non-linear input. This is one of the reasons Apex aim assist is controversial - it provides a precision-at-slow-speed benefit that mouse players cannot replicate without enabling acceleration (which then hurts large flicks).
Beyond the basic Notepad test, several rigorous tests exist:
For most users, the simple visual Notepad test is sufficient verification. Run it once after every Windows major update or driver reinstall.
"Mouse smoothing" (different from acceleration) is interpolation between mouse polls to create smoother on-screen motion. It introduces 5-15 ms latency and reduces input precision. Disable wherever found:
Modern competitive titles do not include user-toggleable smoothing because pros uniformly want it off.
Acceleration interacts with every other aim training decision. Your trainer scores, your DM crosshair placement, your ranked muscle memory - all are built on the input curve in use. Switching the curve mid-improvement-arc resets everything. The reverse is also true: if you have linear muscle memory and switch to acceleration, you lose months of investment.
This is why pros universally lock in their settings early in their career and rarely change. The cumulative investment in trained input behaviour dwarfs the theoretical benefit of any "better" curve. Pick linear, pick early, commit forever.
No. Disable Windows Enhanced Pointer Precision and any in-game acceleration. Linear 1:1 mapping is the pro standard.
Acceleration makes hand-to-aim mapping context-dependent, preventing stable muscle memory.
Tiny minority - mostly Quake-rooted players from the early 2000s with deeply trained accel muscle memory.
Yes. EPP is Windows mouse acceleration. Disable in Control Panel - Mouse - Pointer Options.
Slow swing 10 cm vs fast swing 10 cm. If cursor moves different distances - acceleration is on.
Yes, 10-20 percent for 2-4 weeks. After 5+ weeks, consistency exceeds your prior accel performance.
Raw input bypasses Windows acceleration. It's on by default in every modern competitive FPS - no manual action needed.
Two distinct types of non-linear input exist:
Both are bad for FPS, but negative acceleration is particularly insidious because it manifests during fast flicks - exactly when you need accuracy most. Check for it by comparing slow-swing cursor distance vs fast-swing distance; if fast moves the cursor less, your mouse may be malfunctioning.
For verification:
If you play with friends or a small team, helping teammates disable acceleration may improve overall team performance. Many casual players have Windows EPP enabled without realising. A 60-second walkthrough during a team voice chat - and 4-week adaptation - can lift everyone's aim consistency. Consider sharing this article or the equivalent BlurBusters / Voltaic guides with your team's chat.
In crossplay titles (Apex, Fortnite, Halo, COD), controller players use aim assist - which is essentially server-side acceleration tuned for stick input. Mouse players get pure linear input. The debate over "is aim assist a form of acceleration" is technically yes but practically the systems serve different input devices.
For mouse players in crossplay: stay linear. For controller players: the aim assist tuning is built-in - separate from your mouse acceleration setting which would only matter if you also use a mouse in the same game.
Older sensors (PMW3360 and earlier) had measurable acceleration during very fast movements due to sensor limitations - the optical/laser sensor couldn't track at the high velocities. Modern sensors (PMW3395, PAW3950, Hero 2) are essentially flawless to 750 IPS - well beyond what any human can achieve. If you have a 2018+ premium gaming mouse, sensor-induced pseudo-acceleration is not a concern.
If you have an older mouse and feel "the cursor doesn't go as far on very fast swings", check the sensor's max tracking velocity in the spec sheet. If you're exceeding it, the mouse is artificially capping your input - a form of negative acceleration.
This decision tree covers 99 percent of players. The acceleration question is, in practice, already resolved.