The short answer
Read the event log first, because Windows records why it restarted and that narrows the field immediately. Then check temperatures under load, test the memory, and consider the power supply, which is the most common hardware cause and the one people check last.
A computer that restarts without warning is doing so because something made continued operation impossible: the power dipped, a temperature limit was reached, or the system encountered an error it could not recover from. Each leaves different evidence.
Step 1: read the evidence
Open Event Viewer
Press the Windows key, type Event Viewer, and open it. Go to Windows Logs, then System.
Look for Kernel-Power event 41
This means the system restarted without shutting down cleanly, which is what a power loss or a hard hang produces. It confirms the type of failure but not the cause.
Look for WHEA-Logger events
These are hardware errors, and they often name the component. A WHEA error immediately before a restart is a strong hardware signal.
Look for a bug check code
If a blue screen occurred too quickly to read, the event log records the stop code. Searching that code narrows the cause considerably.
Disable automatic restart on system failure
In System Properties, Advanced, Startup and Recovery, untick Automatically restart. The next failure will show the blue screen and its code instead of restarting instantly.
Step 2: heat
Install a monitoring tool and watch temperatures while the machine is under load. Restarts caused by heat typically happen during games, video encoding or other sustained work, and rarely at idle.
- CPU above 95 degrees under load means the cooler is inadequate, badly mounted, or the paste has failed.
- GPU above 85 degrees is worth investigating, particularly if the card’s fans are audibly struggling.
- Dust in heatsinks and filters is the most common cause and the easiest fix. Compressed air, fans held still while cleaning.
- Case airflow. A machine with one intake and a blocked front panel will cook regardless of cooler quality.
Note that modern processors throttle rather than shut down when hot, so a temperature-related restart usually means something more serious, often the voltage regulators on the motherboard overheating rather than the chip itself.
Step 3: power
The power supply is the most likely hardware cause of random restarts and the least likely to be suspected. Its output degrades with age, and a unit that was adequate when new may no longer handle peak loads, particularly after a graphics card upgrade.
Signs pointing at the power supply:
- Restarts under load rather than at idle, especially at the moment a game loads a new scene.
- Restarts that began after adding a component.
- A unit more than five years old, or an unbranded one supplied with the case.
- No blue screen and no log entry other than Kernel-Power 41, which is the signature of the power simply going away.
Also check the wall side: a failing extension lead, a shared circuit with a heavy appliance, or a loose plug will produce identical symptoms. Plug the computer directly into a wall socket as a test, and note whether restarts coincide with a fridge or heater cycling.
An uninterruptible power supply is a useful diagnostic as well as a protective device. If restarts stop entirely once the computer is on a UPS, the problem was the mains supply rather than the computer.
Step 4: memory
Faulty memory produces restarts, blue screens with varying stop codes, and file corruption. Test it properly:
- Run MemTest86 from a USB stick for at least four passes, which takes hours and is the only reliable test.
- If errors appear, test one module at a time to identify which.
- Check that memory is seated properly and in the correct slots per the motherboard manual.
- If you run an XMP or EXPO memory profile, disable it and test at default speed. Memory overclocking profiles are a very common cause of instability that appears months after a build.
Step 5: storage and drivers
- Check the drive’s health with a SMART tool. A failing boot drive causes hangs and restarts.
- Run sfc /scannow and DISM to repair system files.
- Update or roll back the graphics driver, which is the most common software cause of restarts under load.
- Check for a recently installed program, particularly anything with a kernel driver such as antivirus, VPN or hardware utilities.
- Boot into safe mode and use the machine for a while. Stability in safe mode points firmly at software or drivers.
Disable any overclock, including memory profiles and manufacturer performance modes, before spending time on anything else. An overclock that was stable for a year can become unstable as components age, and testing at stock settings costs nothing.
Step 6: connections
Inside the case, reseat everything: memory, graphics card, power connectors at both ends, and storage cables. Vibration and thermal cycling loosen connections over years. Pay particular attention to the supplementary power connectors on the graphics card, which carry a lot of current and are a known failure point when not fully seated.
Narrowing it efficiently
The pattern tells you most of what you need:
- Only under heavy load: power supply or cooling.
- At idle or randomly: memory, drivers or a failing drive.
- Only after running for hours: heat, or a component that fails when warm.
- Immediately on waking from sleep: drivers or power management.
- Only when a specific device is connected: that device or its driver.
What does Kernel-Power event 41 mean?
That the system was not shut down cleanly. It records the symptom rather than the cause, and it is consistent with a power interruption, a hard hang or a sudden hardware fault.
Can a failing power supply damage other components?
It can, since unstable voltages are hard on everything connected. That is a reason to treat a suspect power supply as urgent rather than something to live with.
How long should I run a memory test?
At least four full passes of MemTest86, which usually means several hours. Short tests miss intermittent faults, and intermittent is exactly what random restarts suggest.
Why does it only restart while gaming?
Games load the processor, graphics card and power supply simultaneously, which exposes marginal power delivery and inadequate cooling. That pattern points at those two before anything else.
Could a Windows update cause this?
Yes, usually through a driver it installs. Check whether restarts began at the same time as an update, and roll back the graphics or chipset driver if so.
The order
Event Viewer first, and disable automatic restart so you can read the stop code. Then match the pattern: under load means power and heat, at idle means memory and drivers. Disable overclocks, clean the dust, test the memory properly, and try the machine on a different mains socket. Reseat connections. Replace the power supply only after the pattern points there, not as a first guess.
