Short answer
Run EventAnalyzer and let it scan the logs for the last 7 or 30 days. Unexpected restarts, unclean shutdowns and power losses show up as critical findings, and for each one it collects the events from the ten minutes before the restart. That is usually where you see whether it was the disk, a driver or the power.
Step by step
-
Run EventAnalyzer as administrator
A single
EventAnalyzer.exe, no installer. It also runs without administrator rights, but then skips the Security log, the crash dumps and some disk data. When hunting for the cause of restarts, it is better to see everything. -
Choose a longer period
Restarts rarely happen every hour. Pick 7 or 30 days so any pattern shows up.
-
Find the restarts among the critical findings
Repeated events are merged into one finding with an occurrence count and the time of the last one, so you see right away how many times the PC restarted and when it last did.
-
Go through the chain of events before the restart
Every crash and restart comes with the events from the ten minutes before. Disk errors, a crashing driver or a hardware error report are the most common clues.
-
Run the recommended fix and scan again
The repair plan is shown in advance with its script and starts with a restore point. After the fix the scan repeats and is compared with the previous one.
What most often causes a restart
| Clue in the logs | Likely cause |
|---|---|
| Power loss with no errors before it | A power cut, a loose cable, a weak or faulty power supply, overheating. |
| Disk errors just before the restart | A failing disk or cable. Check the disk health. |
| A blue screen with a memory dump | A driver or kernel crash, see the stop code. |
| A restart after installing an update | Windows Update, not a fault. |
By default Windows restarts automatically after a crash, so you often never see the blue screen. The memory dump stays behind, though, and EventAnalyzer reads the stop code from it.
Things to keep in mind
- A restart under load (games, rendering) with no errors in the log often means temperature or the power supply. Run a stress test with temperatures.
- Random restarts with a different error each time point to memory, see the RAM test.
- Fixes change the system. Keep the restore point in the plan and read the script.