How to Fix a Computer That Keeps Restarting

What Causes a Computer to Keep Restarting?

A computer keeps restarting because a fault forces the system to reboot, most commonly from overheating, a failing power supply, bad RAM, a driver crash, or a bad Windows update. The reboot is a symptom of a hardware fault or a software crash, not a cause in itself. The four cards below group the usual culprits, ordered roughly by how often each one is to blame.

The timing of the reboot narrows the cause further: a restart under heavy load points to heat or the power supply, while a restart at a fixed point in startup points to a driver, a bad update, or a corrupt system file. A system that reboots before reaching Windows belongs partly to the computer stuck on the boot screen guide, and one that powers off without restarting ties to the computer that turns off by itself guide. A unit that cuts out entirely rather than rebooting is covered in the guide to random PC shutdowns.

How Do You Fix a Computer That Keeps Restarting?

Work the steps in order, starting with the setting that exposes the error and moving to the deeper hardware tests only as each cheaper cause is ruled out. The first step costs nothing and tells you where to spend effort, so do not skip it.

• Disable automatic restart to read the error. Stop Windows rebooting on a crash so the stop code stays on screen. It names the failure class and directs every later step.
• Check CPU and GPU temperatures. Watch them under load with a monitoring tool. A reboot as the CPU nears 100 degrees Celsius confirms thermal shutdown.
• Reseat and test the RAM. Run MemTest86 from a USB drive and test one stick at a time. Bad memory corrupts data and forces reboots.
• Update or roll back drivers. Clean-install the graphics driver, or roll it back if the reboots began right after an update.
• Break the update loop in recovery. If the machine loops before Windows loads, enter the recovery environment and uninstall the latest quality update.
• Scan for malware. Run a full Microsoft Defender scan plus a second-opinion tool, in Safe Mode if needed, to rule out an infection.

Disable Automatic Restart to Read the Error

Disabling automatic restart resolves the inability to read the stop code by stopping Windows from rebooting instantly on a crash. Windows defaults to restarting on a system failure, which hides the blue screen and its stop code before it can be read. Turning the setting off freezes the error on screen for diagnosis. Follow these steps:

1. Open System Properties by searching for ‘View advanced system settings’ in the Start menu, or run sysdm.cpl and open the Advanced tab.
2. Click Settings under Startup and Recovery, then find the System failure section.
3. Clear the ‘Automatically restart’ checkbox, so the system holds the blue screen instead of rebooting.
4. Record the stop code, such as WHEA_UNCORRECTABLE_ERROR or IRQL_NOT_LESS_OR_EQUAL, that the blue screen now displays.

The stop code names the failure class and directs the rest of the diagnosis. Reading and decoding it is the core of the guide to fixing the blue screen of death, which lists what each common code indicates.

Check the CPU and GPU Temperatures

Checking the processor and graphics card temperatures resolves reboots caused by thermal shutdown. A processor that reaches roughly 100 degrees Celsius or a graphics card that exceeds its thermal limit triggers a protective reboot.

Monitoring the temperatures under load confirms or rules out heat. Follow these steps:

1. Install a monitoring tool such as HWiNFO or Core Temp to read the CPU and GPU temperatures in real time.
2. Watch the temperatures under load, since a reboot when the CPU nears 100 degrees Celsius confirms thermal shutdown.
3. Clean dust from the heatsinks and fans with compressed air, because clogged fins block airflow and raise temperatures.
4. Reseat the CPU cooler and reapply thermal paste if temperatures stay high, since dried paste and a loose cooler trap heat.

A system that reboots only under load and runs cool at idle is overheating, so the cooler mounting, the paste, and the case airflow need attention. A system that reboots at idle and stays cool points away from heat toward the power supply, memory, or a driver. The full method for isolating a heat fault appears in the guide to diagnosing PC overheating, and persistent heat that throttles performance also appears in the guide to fixing a slow computer.

Reseat and Test the RAM

Reseating and testing the RAM resolves reboots caused by faulty or unstable memory. Bad memory corrupts data and produces crashes that the system recovers from by rebooting, often with a memory-related stop code or a WHEA_UNCORRECTABLE_ERROR.

Testing the modules isolates a failing stick. Follow these steps:

1. Run Windows Memory Diagnostic by searching for it in the Start menu and restarting to scan the memory for errors.
2. Run MemTest86 from a USB drive for a longer test, and let it complete at least four passes, since errors often appear only on a later pass.
3. Reseat each module until both retention clips snap closed, since a partly seated stick causes intermittent errors.
4. Test one stick at a time to isolate a failing module, then disable any XMP or EXPO profile in BIOS, because that profile is a memory overclock that can destabilize good sticks.

A memory test that reports errors confirms a failing stick that needs replacement, while a clean test with continuing reboots points elsewhere. An unstable XMP or EXPO profile can cause errors on good modules, so disabling it tests the memory at its default speed. Correct module placement for stable operation appears in the guide to installing RAM.

Update or Roll Back Drivers

Updating or rolling back drivers resolves reboots caused by a faulty or incompatible driver. A graphics, chipset, or network driver that is outdated or freshly broken can trigger a stop error and reboot, and the fix depends on whether the problem followed an update. Follow these steps:

1. Identify the driver named in the stop code or the device that fails, focusing on the graphics and chipset drivers first.
2. Clean-install the graphics driver by removing the old one with Display Driver Uninstaller (DDU) in Safe Mode, then installing the latest version from AMD, Nvidia, or Intel.
3. Roll back the driver in Device Manager if the reboots began right after a driver update, reverting to the previous version.
4. Update the chipset and storage drivers from the motherboard maker, since an outdated chipset driver can destabilize the system.

Reboots that began immediately after a driver update point to that driver, so rolling it back tests the link directly. A clean reinstall with DDU removes a corrupted driver and the leftover files that a simple update leaves in place. A driver stop error that also produces a blue screen ties to the blue screen troubleshooting guide.

Break the Update Loop in Recovery

Uninstalling the latest update from the recovery environment resolves a reboot loop caused by a bad Windows update before the desktop ever loads. A faulty cumulative update can leave a machine cycling through the spinning dots, and recent examples on Windows 11 24H2 and 25H2, such as the April 2026 KB5083769 and the February 2026 KB5077181, did exactly that on a subset of PCs. When the loop never reaches Windows you cannot fix it from inside Windows, so you enter the Windows Recovery Environment (WinRE) instead. Follow these steps:

1. Force WinRE to open by holding the power button for about ten seconds to shut the machine off mid-boot, then repeating that twice. On the third start Windows loads the recovery environment instead.
2. Uninstall the latest quality update under Troubleshoot, Advanced options, Uninstall Updates, which removes the most recent cumulative update that broke the boot.
3. Try System Restore from the same Advanced options menu if uninstalling the update does not help, choosing a restore point dated before the trouble began.
4. Run Startup Repair as the next fallback, and if it reports corruption, run SFC and DISM from the WinRE command prompt against the offline Windows drive.

A machine that boots normally after the update is removed had a bad update as the cause, and pausing updates for a week or two lets Microsoft ship a fix before it reinstalls. If the loop continues after the update is gone, the fault is deeper than servicing, so the diagnosis returns to storage or system-file corruption covered in the computer stuck on the boot screen guide.

Scan for Malware

Scanning for malware resolves reboots caused by malicious software destabilizing the system. Some malware forces reboots, corrupts system files, or consumes resources until the system crashes. A full scan with a current tool removes the infection. Follow these steps:

1. Run a full scan with Windows Security, the built-in Microsoft Defender, rather than a quick scan, to check the entire system.
2. Run a second-opinion scan with a reputable on-demand tool such as Malwarebytes to catch what the first scanner misses.
3. Boot into Safe Mode if malware blocks the scan, since Safe Mode loads only essential drivers and services.
4. Remove detected threats and restart, then rescan to confirm the system is clean before resuming normal use.

A system that stops rebooting after malware removal had an infection as the cause. A clean scan with continuing reboots clears malware from the list and returns the diagnosis to hardware or drivers. Running a scan in Safe Mode catches threats that hide while the full operating system runs.

Read the Event Viewer for the Crash Source

Reading the Event Viewer identifies the source of the reboot from the logged error at the time of the crash. Windows records critical errors, including the Kernel-Power event that marks an unexpected shutdown, in the system log. The entries point to hardware, a driver, or a power fault. Follow these steps:

1. Open Event Viewer by searching for it in the Start menu, then expand Windows Logs and select System.
2. Find the Critical and Error entries timestamped at the moment of the reboot to identify the failure.
3. Look for Kernel-Power event ID 41, which marks an unexpected power loss or hard crash consistent with a PSU or heat fault.
4. Note any driver or hardware error in the surrounding entries, which names the component or service that failed.

A Kernel-Power 41 event confirms the system lost power or crashed hard rather than restarting cleanly. An Event 41 with no recorded bug check leans toward the power supply, while one paired with a stop code leans toward heat, memory, or a driver. The same logs help diagnose a system that powers off entirely in the computer that turns off by itself guide.

Test the Power Supply Under Load

Testing the power supply under load resolves reboots caused by a unit that cannot hold its rails when the system draws heavy current. A failing or undersized power supply drops a rail when power draw spikes, cutting power long enough to force a restart, often during gaming or rendering. Follow these steps:

1. Reseat the 24-pin and 8-pin EPS connectors at the motherboard, since a loose plug drops power under load.
2. Test the power supply with a PSU tester or multimeter to confirm each rail holds its voltage within tolerance.
3. Swap in a known-good power supply of adequate wattage to confirm whether the original unit causes the reboots.
4. Confirm the wattage is sufficient for the processor and graphics card, keeping a 20 to 30 percent buffer over the calculated load to absorb the transient spikes a graphics card produces.

A system that stops rebooting with a known-good power supply had a failing or undersized unit. How the power supply delivers and protects its rails appears in the explanation of how power supplies work, and sizing the unit for the load appears in the guide to PSU wattage.

Last Thoughts on a Computer That Keeps Restarting

A computer that keeps restarting is reacting to a fault, so the fix begins by exposing the error rather than guessing at parts. Disable automatic restart to read the stop code, then work the causes in order of likelihood: check temperatures, test the RAM, update or roll back drivers, break any update loop from the recovery environment, scan for malware, and test the power supply under load. The timing of the reboot and the stop code do most of the narrowing, so a restart under load sends you to heat and the power supply, while a loop before the desktop sends you to a bad update or corrupt files.

Reading the Event Viewer ties the loose ends together, since a Kernel-Power 41 entry separates a hard power loss from a clean restart. Readers can continue with the fix for the blue screen of death, the fix for a computer that turns off by itself, the guide to random PC shutdowns, or the hub of common PC problems for related faults.

Key Takeaways:

• Disable automatic restart first, so Windows holds the blue screen and reveals the stop code that directs the diagnosis.
• Use the timing as a clue: a reboot under load points to heat or the power supply, an idle reboot to memory or a driver, and a loop before the desktop to a bad update.
• Check CPU and GPU temperatures under load, since a reboot near 100 degrees Celsius is thermal shutdown from dust, dried paste, or a loose cooler.
• Test the RAM with MemTest86 for at least four passes, and disable XMP or EXPO to rule out an unstable memory overclock.
• Break a Windows update loop by entering the recovery environment and uninstalling the latest quality update, then pausing updates.
• Read the Event Viewer for Kernel-Power event 41, and test the power supply under load, keeping a 20 to 30 percent wattage buffer.

Frequently Asked Questions (FAQs)