Why Your PC Restarts by Itself

Quick Answer: Six Causes by Frequency

Automatic restarts have six main causes. In service statistics, they rank roughly as follows: the power supply, overheating, memory, drivers, storage, and finally system settings or third-party software.

• The power supply tops the list, especially when restarts follow load. Aging capacitors stop handling the transient spikes produced by a modern graphics card.
• Overheating triggers protection that shuts the machine down before components are damaged. The defining sign is a crash after 10–20 minutes rather than immediately.
• Memory creates a random pattern: restarts under load, at idle, and while copying files.
• Drivers usually cause a blue screen rather than an immediate reboot, which is an important distinction.
• Storage tends to cause a freeze followed by a reboot, together with log errors and long loading times.
• Settings and software include automatic restarts after updates, unstable overclocking, and conflicting applications.

First: Record the Pattern

Before checking anything, answer four questions. They narrow the search more effectively than any benchmark.

• When exactly does it happen? Under load in a game, at idle, when waking from sleep, when connecting a device, or when launching a demanding application.
• Is there an error screen? A blue screen with a stop code, or an immediate loss of display with no message. These symptoms are fundamentally different.
• What changed beforehand? A new graphics card, a driver or Windows update, moving or cleaning the PC, a power incident, or a different power strip.
• How often does it happen? Once an hour under load, once a day at idle, or only during one specific action.

Correlation with load is the most useful clue. If the restarts happen only when the machine is running at full power, the power supply and cooling move to the top of the list. If they are random and unrelated to load, start with memory and storage.

Blue Screen or Instant Shutdown: How to Tell and Why It Matters

The difference between these two scenarios determines the entire diagnostic path.

A blue screen followed by a restart means the Windows kernel detected an error, wrote a dump file, and restarted the system in an orderly way. This points to the software path: a driver, memory, or system files. If automatic restart is enabled, the blue screen may flash so quickly that it looks like a normal reboot.

An immediate shutdown without a message, where the display disappears and the machine starts booting again, means a hard reset: the operating system had no opportunity to record anything. This points to power, thermal protection, or a hardware fault.

To distinguish them conclusively, disable automatic restart: open System Properties, then Advanced, then Startup and Recovery, and clear “Automatically restart.” The next blue screen will remain visible with its stop code. If the machine still dies silently, it is a hard reset and the cause is hardware-related.

Event Viewer: What to Look For

Event Viewer stores the history of every crash, and event IDs reveal what type of failure occurred. Open eventvwr.msc, then go to Windows Logs and System.

The most common and most ambiguous case is Kernel-Power event 41 without a matching BugCheck. It directly confirms that the machine died instantly rather than at Windows’ request. The only remaining question is what interrupted power: the PSU itself, thermal protection, an unstable overclock, or a poor connection.

If a BugCheck entry with a stop code is present, diagnostics follow a different path described in our guide to Windows blue screens: how to read the stop code and identify the cause.

Cause 1: Power Supply

The power supply is the first suspect whenever restarts follow load. The mechanism is simple: during a power spike, rail voltage drops below the permitted threshold and the system resets before it can record anything.

Signs that point to the power supply:

• Crashes occur only in games, rendering, or stress tests, and never at idle.
• The problem began after installing a more powerful graphics card.
• The PSU is more than six or seven years old, even if its rated wattage is theoretically sufficient. The UAE climate accelerates capacitor aging, so this threshold arrives sooner than in temperate markets.
• The monitor flickers or changes brightness when load rises.
• The PSU produces clicking, whining, or an unusual smell.
• The unit does not meet ATX 3.0 or 3.1 requirements for a modern GPU. New cards produce short spikes well above rated consumption, and a PSU not designed for them may trigger protection.

What you can check without replacing the PSU: make sure every graphics-card power connector uses a separate cable rather than a daisy-chain, remove questionable adapters and extension leads, and verify that the mains supply is stable. Limiting the GPU power target to 80–90% is a useful test: if the crashes stop, power delivery is confirmed as the cause.

Conclusive proof requires installing a known-good power supply with sufficient capacity. This is the main reason PSU faults are difficult to diagnose at home: most users do not have a spare unit.

Cause 2: Overheating

Overheating causes an emergency shutdown rather than a true restart: protection trips before components are damaged. The characteristic sign is timing—the PC works normally for 10–20 minutes and then dies.

What to check:

• CPU and GPU temperatures under sustained load, not only during the first few minutes. Stabilized values are what matter.
• Fan speeds: fans suddenly reaching maximum speed when the system used to be quiet indicate degraded heat transfer.
• Dust in heatsinks and filters. This is the leading cause in any machine older than a year. In the UAE, ambient heat and fine dust clog filters faster, making six to eight months a realistic service interval instead of twelve.
• The condition of thermal interfaces. Paste dries out, while graphics-card thermal pads lose conductivity after three to four years.
• Pump operation in liquid-cooled systems: a stopped pump causes a very rapid temperature rise.

If temperatures are near their limits, technical service with cleaning and thermal-interface replacement costs far less than replacing components.

Cause 3: Memory

Memory produces the most random pattern: crashes under load, at idle, and while copying large files, without a link to one specific action. Blue screens with changing stop codes are also common.

Diagnostic order:

• Disable the memory overclocking profile (XMP or EXPO) and test at standard speed. An unstable profile is very common and reflects platform compatibility rather than necessarily defective memory.
• Run MemTest86 from bootable media for several passes. The built-in Windows test is only a rough check.
• Test one module at a time when several are installed, so the faulty one can be isolated accurately.
• Move the modules to other slots: sometimes the slot, not the memory, is faulty.
• Check that every module is fully seated, especially if the computer was moved recently.

A single test error is a fault, not statistical noise. One 2026 detail is worth noting: because of the DRAM shortage, working modules you remove retain real resale value, so the net cost of a memory replacement is lower than it first appears.

Cause 4: Drivers and Updates

Drivers normally cause a blue screen rather than an instant shutdown, so this section applies when Event Viewer contains a BugCheck entry.

• The graphics driver is the most frequent source. Reinstall it after fully removing the previous version rather than installing over it; follow the complete driver removal procedure.
• Chipset and storage-controller drivers are the second most common group, especially after a Windows update.
• A recent system update. If crashes started immediately afterward, rolling it back provides a quick answer.
• The network adapter is more likely to cause disconnects and freezes than restarts.
• Anti-cheat services and security-software filter drivers can conflict with one another and with the operating system.

Cause 5: Storage

A failing drive usually appears as a freeze followed by a reboot rather than an immediate reset. Warning signs include long load times, the system freezing for several seconds, and Disk or Ntfs errors in Event Viewer.

What to check:

• SMART attributes: reallocated sectors, read errors, and consumed write endurance.
• Free space on the system drive: an almost-full drive can make the system unstable.
• SATA and power cables; for NVMe drives, verify that the device is firmly seated in the slot.
• The file system with chkdsk C: /f /r.

Cause 6: System Settings and Third-Party Software

This group is the least dangerous and should be checked first because it costs nothing.

• Automatic restarts after Windows updates. If the PC reboots at a fixed time without load, start here and configure Active Hours in Windows Update.
• Fast Startup can conflict with drivers and cause failures during shutdown or wake. Disable it in Power Options.
• CPU and GPU overclocking. Remove every unstable overclock before diagnostics, otherwise all other tests become unreliable.
• Competing monitoring and tuning utilities: two applications controlling fan curves at the same time are a classic source of instability.
• Malware and hidden cryptocurrency miners create load, heat, and restarts. Check with a security scan and by monitoring idle utilization.

Scenario and Suspect

Self-Diagnostic Order

• Disable automatic restart so that you can see whether a blue screen appears.
• Read Event Viewer: Kernel-Power 41 without BugCheck points to hardware, while BugCheck 1001 points to software.
• Remove all overclocking, including memory profiles.
• Check temperatures under sustained load.
• Run MemTest86, testing one module at a time if several are installed.
• Check SMART data and free space on every drive.
• Reinstall the graphics driver after completely removing the old version.
• Limit the GPU power target to 80–90%. If the crashes stop, power delivery is the cause.
• Inspect power connectors, avoid daisy-chaining, and eliminate questionable adapters.

If all nine steps fail to identify the cause, only the components that require known-good substitutes remain: the power supply, motherboard, and processor.

When to Use Professional Service

There are situations in which trial and error at home costs more than professional diagnostics:

• WHEA-Logger entries are present, meaning the hardware itself reported an error.
• MemTest86 reports errors, but several modules are installed and the faulty one must be isolated.
• Crashes happen only under load, the PSU is suspected, and no replacement unit is available.
• The machine dies instantly and shows no blue screen even after automatic restart is disabled.
• The power supply produces any smell, whine, or clicking. Stop using the machine immediately.
• The machine is under warranty, because opening it yourself may void coverage.

Technical service includes memory and storage checks, sustained-load stability testing, and replacement of suspect components with known-good parts. The Upgrade Center page explains how systems are accepted in Dubai, including courier collection. This is where access to a parts inventory can replace a week of experiments.

Recommended