PC Keeps Restarting? Fix the Random Reboots (2026)

By LK Wood IV · 2026-07-04 · ~10 min read · St. Louis County, MO

Some retailer links below are Amazon affiliate links; if you buy through them TechFuel HQ may earn a small commission at no extra cost to you. As an Amazon Associate I earn from qualifying purchases. Commissions never influence what gets recommended — see our disclosure.

A PC that keeps restarting on its own is one of the most stressful problems to diagnose, because the reboot wipes the evidence before you can read it. The fix starts with a setting most guides skip: turn off Windows’ automatic restart so the next crash actually shows you the error. From there it’s a short, ordered checklist — heat, power, RAM, drivers — and the pattern of when it restarts tells you where to look.

Step 1: Make Windows show you the error (do this first)

By default, Windows instantly reboots on a system crash, so you never see the blue screen. Turn that off so the next crash leaves a stop code on screen:

1. Press Windows Key + R, type sysdm.cpl, and press Enter.
2. Go to the Advanced tab → under Startup and Recovery, click Settings.
3. Under System failure, uncheck “Automatically restart”, and click OK.

Now, the next time it crashes, you’ll get a blue screen with a stop code instead of an instant reboot. Write it down — a code like WHEA_UNCORRECTABLE_ERROR points at hardware or power, while a named driver file (nvlddmkm.sys, for example) points at the GPU driver. That one clue turns guesswork into a targeted fix.

If the PC restarts with no blue screen at all — an instant, total cut like someone pulled the plug — that’s a strong sign of a power problem (PSU or a loose cable), because a true power loss happens faster than Windows can even draw a BSOD. And if it doesn’t come back on its own at all — it powers off and stays off until you press the button — that’s a sudden shutdown rather than a restart, with its own checklist: work our computer randomly turns off guide.

If instead of rebooting the system locks up and freezes — the screen stays frozen, the mouse won’t move, and it never restarts on its own — that’s a different failure with a different checklist; work our computer keeps freezing guide instead.

Five of the most-searched stop codes have their own step-by-step deep dives: DPC_WATCHDOG_VIOLATION (a driver stalling the CPU — usually storage), KERNEL_SECURITY_CHECK_FAILURE (memory corruption — test RAM first), CLOCK_WATCHDOG_TIMEOUT (a CPU core that stopped responding — revert the overclock first), PAGE_FAULT_IN_NONPAGED_AREA (missing memory — test RAM, then storage), and SYSTEM_THREAD_EXCEPTION_NOT_HANDLED (a bad driver, usually named right on the blue screen). If your blue screen shows any of them, start there.

Read the pattern

When it restarts narrows it down fast:

• Only under load / while gaming → heat or power (jump to Overheating and PSU).
• Random, including at idle or light use → RAM, driver, PSU, or an unstable overclock. When it’s truly random and the stop code keeps changing, use the isolation build below.
• In a loop before Windows loads → software (bad update, corrupt files) or a hardware failure (see the boot-loop FAQ above).
• Only when you shut down — it reboots instead of powering off, or won’t stay off → that’s usually not a crash at all; see restarts right after shutdown.

Step 2: Overheating

A CPU or GPU that hits its thermal limit reboots the system to protect itself, and this is the #1 cause of load-triggered restarts. If the CPU is the culprit, the why is my CPU overheating guide sorts the cause by when it happens - a dead pump, dried paste, or a loose cooler each fix differently. If the restarts hit hardest in games, suspect the graphics card and confirm the GPU is really overheating with the core-vs-hotspot check before you touch anything - a card that only looks hot needs a different fix.

• Check temperatures under load with HWiNFO or HWMonitor. Our safe CPU temperature guide covers what’s normal (Ryzen to ~95°C, X3D ~89°C, Intel ~100°C) and when heat is actually the problem — a chip parking at its limit and rebooting means cooling can’t keep up.
• Clear dust from fans, heatsinks, and vents with a can of compressed air, and confirm every fan spins (a failed AIO pump or CPU fan causes fast thermal shutdowns).
• Reapply thermal paste if it’s a few years old — dried paste is a common cause of a system that used to run cool and now overheats. An inexpensive tube of Arctic MX-4 is enough for several repastes and often drops load temps several degrees.

Step 3: Power supply

The PSU is the most under-suspected hardware cause of restarts, especially the load-related and no-BSOD kind. A supply that’s failing, undersized for your GPU, or connected with a loose cable can’t hold the load and drops the system.

• Reseat every power cable — the 24-pin, the CPU (EPS) connector, and every GPU power cable, including all pins on a 12VHPWR/12V-2x6 connector, which is notorious for partial seating.
• Rule out undersizing: a new, power-hungry GPU on an old PSU is a classic restart cause. Run your parts through the PSU wattage calculator to see whether your supply has the headroom — if you’re near or over its rating under load, that’s your answer.
• If temps are healthy and it still restarts under load, a known-good, adequately-rated PSU is the next test. Size it first, then buy — don’t guess on wattage.

If power looks like the culprit and you want to confirm the PSU is failing rather than just undersized, our deep dive covers the tell-tale symptoms, why the paperclip test can give a false all-clear, and how to read the 12V rail under load: PSU Failure Symptoms: Is It Really the Power Supply?

Step 4: RAM

Faulty or unstable memory causes random restarts and blue screens.

• Test it: run the built-in Windows Memory Diagnostic (Windows Key + R → mdsched.exe), or MemTest86 from a USB stick for a thorough overnight pass.
• Reseat and isolate: reseat both sticks, then test one stick at a time to find a bad module.
• Check your memory speed: an unstable EXPO/XMP profile can restart a system that’s otherwise fine. Set memory to its default JEDEC speed to test; if the restarts stop, the kit or memory controller isn’t holding the rated speed.

Step 5: Drivers and overclocks

• Graphics driver: a crashing GPU driver can take the whole system down. If your blue-screen names a driver like nvlddmkm.sys, do a clean reinstall — the full process is in the GPU driver crash fix guide.
• Revert overclocks/undervolts: set any CPU or GPU tuning (MSI Afterburner, Ryzen Master, BIOS PBO/curve) back to stock. An overclock that’s slightly unstable, or an undervolt that’s too aggressive, restarts the system under load exactly like a hardware fault.

Step 6: Software (for boot loops and post-update restarts)

If the trouble is a reboot loop or started right after an update:

• Windows Recovery: force-power-off during boot three times to trigger Automatic Repair, then try Startup Repair, System Restore, or Uninstall latest updates.
• Safe Mode to roll back a bad driver or uninstall recently installed software.
• Scan for malware, and run sfc /scannow then DISM /Online /Cleanup-Image /RestoreHealth to repair corrupt system files.

When it's random: the isolation build

When a restart only happens under load, or the blue screen names a part, you follow the step it points to. But some restarts are genuinely random — different stop codes each time, no pattern, or no blue screen at all — and chasing those by swapping parts wastes money. The reliable move is the opposite of adding parts: strip the machine to a configuration you know is stable, then add each variable back one at a time until the restart returns. Whatever you re-enabled last is the cause.

The two variables that cause the most “random” restarts are the ones people forget they turned on: the memory’s EXPO/XMP profile and any CPU or GPU overclock, PBO, or undervolt. Both run the hardware past its guaranteed-stable defaults, and both can behave for weeks and then drop the system under the wrong workload. Turning them off costs nothing and clears them in one test — which is why the isolation build starts there, not at the parts store.

Restarts right after you shut down (or won't stay off)

There’s one “restart” that isn’t a crash at all: you click Shut down, the PC powers off for a second, then boots straight back up — or it never fully powers off. That’s a shutdown problem, not a stability one, and it has its own short list.

• Turn off Fast Startup first. Fast Startup (Windows’ hybrid shutdown) saves the kernel to disk instead of powering all the way down, and on a lot of systems it’s the direct cause of a reboot-instead-of-shutdown. Open Control Panel → Power Options → “Choose what the power buttons do” → “Change settings that are currently unavailable”, then uncheck “Turn on fast startup” and save. Forcing a full cold shutdown every time fixes the majority of these cases.
• Stop a device from waking it. A network adapter or mouse allowed to wake the machine can pull it back on the instant it powers off. In Device Manager, open the network adapter (and the mouse) → Power Management tab → uncheck “Allow this device to wake the computer.” From a terminal you can list armed wake sources with powercfg /waketimers and see what last woke it with powercfg /lastwake.
• Rule out a pending update. A Windows Update mid-install reboots rather than shuts down once — let it finish, then try again.
• A dirty previous shutdown (a hard power-off or a crash) can make the next shutdown behave oddly a single time; if it happened only once, it’s likely nothing.

If it still reboots on shutdown after all of that, treat it like a random restart and work the stability steps above — a marginal PSU can occasionally show up this way too. And if the machine will not power on at all rather than rebooting, that is a different fault tree — start with why a PC power button stops working.

The fix order, in one list

1. Uncheck “Automatically restart” (sysdm.cpl → Advanced → Startup and Recovery) so you can read the stop code.
2. Read the pattern — under-load = heat/power; random = RAM/driver/PSU/overclock; boot-loop = software/hardware failure.
3. Overheating — temps, dust, fans, thermal paste.
4. PSU — reseat cables, check wattage headroom, then swap if load-related.
5. RAM — mdsched/MemTest86, reseat, one stick at a time, test at JEDEC speed.
6. Drivers/overclocks — clean-reinstall the GPU driver, revert all tuning to stock. Still random? Run the isolation build — strip to a stable floor, add one variable back at a time.
7. Software — Automatic Repair, Safe Mode, malware scan, sfc/DISM for boot loops.

Work it in order and the stop code from step 1 usually points you straight to the right step — no parts-cannon required. Two cases sit outside this flow: a genuinely random restart is fastest to pin with the isolation build, and a reboot that only happens when you try to shut down is a Fast Startup problem, not a crash.

Sources

• The diagnostic and fixes use Microsoft’s built-in Windows tools — disable automatic restart (sysdm.cpl → Advanced → Startup and Recovery), the Windows Memory Diagnostic (mdsched.exe), Automatic Repair / Safe Mode, and sfc /scannow + DISM /Online /Cleanup-Image /RestoreHealth — all standard and self-verifiable on any Windows 10/11 PC. The ordered hardware causes (overheating, PSU, RAM, driver, overclock) are cross-corroborated across community and vendor troubleshooting guides.
• Thermal limits and what “too hot” actually is: our safe CPU temperature guide (first-party). Driver-crash clean reinstall: our GPU driver crash fix guide (first-party).
• Blue-screen stop codes map to Microsoft’s Bug Check Code Reference (WHEA_UNCORRECTABLE_ERROR = fatal hardware error; CLOCK_WATCHDOG_TIMEOUT = an unresponsive CPU core; VIDEO_TDR_FAILURE = a display-driver timeout). Kernel-Power event ID 41 is Microsoft’s marker that the PC rebooted without a clean shutdown — a symptom of a power loss or hard crash, not a cause (Microsoft Q&A, corroborated on Tom’s Hardware and Linus Tech Tips forums).
• The restart-instead-of-shutdown fix (disable Fast Startup, then check device wake permissions) is the standard, widely-corroborated remedy — see ASUS official support and The Windows Club.

Menu names below match 2026 Windows builds; Microsoft moves Settings pages between versions. Email hello@techfuelhq.com if a step does not line up.