How to Fix Game Stuttering on Windows 11 (2026 Guide)

Stuttering can be difficult to diagnose from average FPS alone, because the number everyone looks at is the wrong number. A game running at a steady 165 FPS and a game averaging 165 FPS while dropping a 40 ms frame every two seconds report the same average. The same average FPS can describe different frame-time distributions and perceived smoothness.

This guide is the diagnosis-first version, because ten tweaks applied blindly to an unknown cause is how people end up with a heavily modified Windows install that still stutters. Work through it in order.

Step 1: Understand what you are actually measuring

Frametime, not frame rate

Frame rate is frames divided by seconds — an average over a window. Frametime is how long each individual frame took, in milliseconds. At 100 FPS a perfectly smooth game delivers a 10 ms frame, then another, then another. Stutter is what you see when that sequence becomes 10, 10, 47, 10, 10, 38. The average barely moves; your eyes and hands notice immediately.

This changes what counts as a fix. Anything that raises your average FPS while leaving the spikes in place has not helped. Anything that flattens the spikes has, even if the average drops slightly.

Tools worth installing

• CapFrameX — captures and plots frametime data, with 1% and 0.2% low percentiles. Those percentile lows describe how the game feels. Record a two-minute session before and after each change.
• PresentMon — Intel's frame capture tool, the engine underneath many overlays. Use it when you want raw data on how frames move through the presentation pipeline.
• HWiNFO64 — sensor monitoring. Run it in sensors-only mode with logging while you play, then check CPU and GPU temperatures, clock speeds over time, and the thermal or power limit indicators your platform exposes.

Capture one baseline before you change anything. Without it, every subsequent measurement is an opinion.

Three patterns and what they mean

1. Stutters clustered in the first minutes or in new areas, smooth on a second pass — may suggest shader compilation.
2. Smooth for five to fifteen minutes, then progressively worse, recovering after the machine cools — may suggest a thermal limit.
3. Random spikes with no pattern, sometimes on the desktop too — may suggest background processes, storage or drivers.

These patterns suggest possibilities but do not identify a cause by themselves. Confirm with shader activity, temperature and clock logs, processes, storage and drivers.

Step 2: Shader compilation stutter

A frequent suspect in modern games, and not a fault in your PC. Games built on Unreal Engine and similar pipelines compile shaders on your machine, for your exact GPU and driver combination. If rendering must wait for shader compilation, a frame-time spike may occur. Compilation behavior depends on the game.

How to recognise it: stutters happen the first time you enter an area, see a new effect, or encounter an ability you have not seen this session. If a second pass is smoother, shader compilation or asset caching may be worth investigating. If the game offers shader pre-compilation at startup, let it finish — Skipping pre-compilation may leave compilation work for gameplay, depending on how the game handles shaders.

Practical mitigations: let the precompile step finish, keep the driver shader cache size generous rather than disabled, and avoid cache wipes as routine maintenance. Our shader compilation stutter guide walks through it per engine, and the NVIDIA Control Panel guide covers the cache setting.

Step 3: Thermal throttling

Another suspect worth checking, and one people often tweak around instead of fixing. When a CPU or GPU reaches its thermal limit it reduces clocks to protect itself. Thermal throttling may lower frame rates or affect frame-time consistency. Check whether clock changes coincide with measured spikes.

The signature is timing: the first few minutes are fine, then it degrades, and after a break it is fine again. Confirm it in HWiNFO by logging temperatures and clocks through a full session and looking for the moment clocks fall while load stays high.

What to do, in order: clear dust from intakes, fans and heatsinks; make sure intake and exhaust are not blocked; on a laptop, lift the chassis off the desk so the underside intake is clear; check that fan curves are not set to near-silent; and if the machine is several years old, consider that thermal paste degrades. Gaming laptops are especially prone to this because their cooling headroom is small by design.

Step 4: Background apps, overlays and Discord

Overlays interact with game presentation in different ways. Test the active ones individually. Test active overlays individually; their number alone does not establish whether they cause a problem. The usual suspects are Discord, Steam, the NVIDIA app, MSI Afterburner and RivaTuner, Xbox Game Bar, and whatever your peripheral vendor installed.

Discord deserves its own mention because it is on nearly every gaming PC and has two independent things that can cause stutter: hardware acceleration and the in-game overlay. Turn off the overlay first, test, then hardware acceleration, test again. Our Discord FPS drops guide has the exact settings.

Beyond overlays, look at what consumes CPU during a match: open Task Manager, sort by CPU, and play windowed for a minute. Browser tabs with video, cloud sync clients indexing a large folder, and scheduled antivirus scans are all common. If one process sits high throughout, follow high CPU usage while gaming. If dozens of things start with Windows, slow Windows startup and our bloatware guide cover the cleanup safely.

Step 5: Power plan and CPU power behaviour

A power plan tuned for efficiency parks cores and drops clocks aggressively between bursts of load. In a game alternating between light and heavy frames, clock transitions may affect frame times on some systems. Compare power plans using the same repeatable scenario.

On a desktop, Balanced is usually fine on modern platforms, but High Performance is a reasonable test to isolate the variable. On a laptop, compare the available power modes and check whether battery operation imposes power limits. The effect depends on the model, firmware and workload. Check the manufacturer's own performance mode too, since vendor utilities frequently override the Windows plan entirely.

Test it properly: capture a frametime graph on your current plan, switch, capture again. If the graph does not change, put it back. Our power plan guide covers where the differences actually matter.

Step 6: HAGS and MPO

Two display-stack features that help on some systems and cause visible stutter on others. Both are worth testing precisely because their effect is hardware dependent, and neither has a universally correct setting.

Hardware-Accelerated GPU Scheduling

HAGS moves some scheduling work from the CPU to the GPU. HAGS may improve, worsen or leave performance unchanged, depending on the game, GPU and driver. Compare both states with repeatable frame-time captures. It lives in Settings, System, Display, Graphics, under default graphics settings, and needs a restart. Test both states with a frametime capture rather than by feel — this is exactly the kind of setting where expectation shapes perception.

Multi-Plane Overlay

MPO lets the display engine present multiple layers without going through the full desktop composition path. On certain driver and monitor combinations, particularly multi-monitor setups with mixed refresh rates, it has been associated with flickering and stutter. A simpler first test on a mixed-refresh setup: temporarily unplug the second monitor and see whether the stutter disappears.

Step 7: Storage and disk saturation

Open-world and streaming-heavy games load assets continuously. If asset loading is delayed by storage activity, frame-time spikes may occur. Check whether disk activity coincides with the measured hitches. Some games may encounter storage-related hitches on a mechanical drive. Check disk activity and the game's storage requirements.

Watch disk activity in Task Manager while playing. High drive active time is a clue, not proof of the game's bottleneck. Check whether it coincides with frame-time spikes and which process is using the drive. Check free space too — Low free space may affect some SSD workloads; the effect depends on the drive and workload. Also check whether something else is hammering the disk: search indexing after a large file move, a cloud sync client, or a backup job. Full detail is in 100% disk usage while gaming.

Step 8: Drivers, especially after a Windows update

A very common scenario: everything was fine, Windows installed an update, now the game stutters. Windows Update sometimes replaces a GPU driver with an older WHQL version, and feature updates occasionally reset graphics settings you had configured.

Reinstall the current driver from NVIDIA, AMD or Intel directly, choosing the clean install option so the previous configuration is removed rather than merged. If the stutter started immediately after a driver update rather than a Windows one, try the previous driver branch — new drivers occasionally regress for specific games. Both paths are covered in game stuttering after a Windows update and Windows 11 25H2 gaming performance.

Step 9: RAM and XMP/EXPO stability

Memory instability can produce errors, crashes or other symptoms. Investigate it when supported by timing, test errors or changes to memory settings. If you enabled XMP or EXPO and stutter appeared around the same time, that is a strong hint.

Check two things. First, that the profile is actually applied — many boards run memory at a conservative JEDEC speed until you enable it, which may affect performance depending on the workload and memory configuration but worth knowing. Second, stability: run an extended memory test, and if errors appear, back the profile down a step or run at stock and re-test. Mismatched kits and four-stick configurations are more sensitive than two matched sticks.

Insufficient RAM is a separate issue with a similar feel: Memory pressure can lead to paging and may affect frame times. Check memory and storage activity alongside the stutters. Watch memory usage during a session before blaming timings.

Step 10: Mouse polling and input-side stutter

Checked last here, but the symptom is distinctive: stutter correlated with mouse movement, worse when you flick, and often visible on the desktop as well as in games. Very high polling rates can generate enough interrupts to cost measurable CPU time on some systems, and certain USB controller and hub combinations handle it badly.

Test by dropping the polling rate one step, plugging the mouse into a rear motherboard port rather than a hub or front panel, and trying a different USB controller if your board has more than one. If the desktop cursor stutters too, the problem is not your game. See mouse stutter on Windows 11, and reduce input lag if the issue is latency rather than stutter.

Game-specific notes

Some titles have their own patterns. Apex Legends and Marvel Rivals lean heavily on streaming and shader compilation; Fortnite behaves differently by rendering mode; FiveM adds a large modded asset load; Roblox workloads vary by experience; measure CPU activity and background contention in the experience you play. Check for a known quirk before blaming your hardware.

The discipline that makes this work

Change one thing. Re-measure. Keep or revert. Move on.

That is the entire method, and it is the part almost everyone skips. Applying eight tweaks at once teaches you nothing, and when the stutter returns in three weeks you are eight changes away from a known-good state. Keep a note of what you changed and how to undo it — or use a tool that keeps that record for you.