A blue screen on a custom PC rarely means a single broken part. It means Windows hit an error it could not recover from and forced a stop to protect your hardware and data. Because custom builds mix parts, BIOS versions, and drivers from different vendors, the root cause is often a mismatch or instability rather than outright failure. This guide gives you a repeatable way to identify the cause, test it safely, and reverse any change you make.
Before you start, enable crash logging so you have data to work with. Press Win + R, type sysdm.cpl, press Enter, go to Advanced > Startup and Recovery > Settings and confirm Write debugging information is set to Automatic memory dump and Automatically restart is unchecked. Your dump files will then appear in C:\Windows\Minidump and C:\Windows\MEMORY.DMP. You can review them with Event Viewer at Event Viewer > Windows Logs > System filtered for source BugCheck.
Most Common Causes
Use this table to match what you see to the most probable cause. Likelihood reflects patterns seen in custom builds where memory, power, and driver configuration vary.
| Cause | Symptom | Likelihood |
|---|---|---|
| Unstable RAM / Incorrect XMP/EXPO Profile | Random codes like MEMORY_MANAGEMENT, IRQL_NOT_LESS_OR_EQUAL, PAGE_FAULT_IN_NONPAGED_AREA; crashes under load and at idle | Very High |
| Driver Conflict or Corrupt Driver | Crashes after GPU, chipset, or network driver update; often DRIVER_IRQL_NOT_LESS_OR_EQUAL, VIDEO_TDR_FAILURE, DPC_WATCHDOG_VIOLATION | Very High |
| BIOS / Firmware Incompatibility | Blue screens after new CPU/RAM install, failure to boot with EXPO/XMP enabled, WHEA_UNCORRECTABLE_ERROR | High |
| Overheating CPU or GPU | Crashes only during gaming or stress tests, fans at full speed, THERMAL_EVENT or CLOCK_WATCHDOG_TIMEOUT before shutdown | High |
| Power Supply Instability | Reboots and blue screens under sudden load, KERNEL_POWER 41 in Event Viewer, crashes when GPU spikes | Medium-High |
| Storage or File System Corruption | KERNEL_DATA_INPAGE_ERROR, UNEXPECTED_KERNEL_MODE_TRAP, slow boots, chkdsk prompts | Medium |
| Corrupt Windows System Files | Crashes after failed update or improper shutdown, SFC finds integrity violations | Medium |
| Faulty Peripherals or PCIe Device | SYSTEM_THREAD_EXCEPTION_NOT_HANDLED pointing to UsbHub3.sys or similar, crashes when device is plugged in | Low-Medium |
Diagnose Step by Step
Work in order. Do not change multiple settings at once or you will not know what fixed it. As of early 2026, Windows 11 23H2 and 24H2 use the same core diagnostic tools, so these paths apply to both.
1. Capture the Exact Stop Code and File
When the blue screen appears, photograph the screen. Note the stop code at the bottom and any file named, such as nvlddmkm.sys, amdkmdag.sys, or ntfs.sys. After reboot, open Event Viewer > Windows Logs > System and look for BugCheck events in the last hour. Double-click the event to see BugCheckCode and parameters. Also check C:\Windows\Minidump for .dmp files dated to the crash.
2. Boot to Safe Mode and Check Recency
Press Win + I > System > Recovery > Advanced startup > Restart now, then choose Troubleshoot > Advanced options > Startup Settings > Restart and press 5 for Safe Mode with Networking. If crashes stop in Safe Mode, the cause is almost certainly a third-party driver or startup software. Ask what changed in the last 10 days: new driver, Windows Update at Settings > Windows Update > Update history, new RAM, or BIOS flash.
3. Test Memory Before Anything Else
Unstable memory causes misleading codes. Press Win + R, type mdsched.exe, choose Restart now and check for problems for a quick scan, then run a thorough test with MemTest86 from a USB if errors persist. In BIOS, note your current DRAM settings under AI Tweaker / OC / Extreme Tweaker > DRAM Status or Settings > Overclocking > DRAM Configuration. Record frequency, voltage, and timings before changing anything.
4. Check Temperatures and Voltages
Install HWiNFO64 and monitor Sensors during idle and load. For desktops, expect CPU idle 30 to 45 degrees Celsius, load under 85 to 90 degrees Celsius for most modern chips, and GPU hotspot under 100 degrees Celsius. In BIOS, check Hardware Monitor / Monitor for +12V rail within 11.4V to 12.6V, +5V within 4.75V to 5.25V, and +3.3V within 3.135V to 3.465V. Sustained deviation or rapid voltage swing under load points to cooling or power issues.
5. Verify Drivers and Storage Health
Press Win + X > Device Manager and expand Display adapters and IDE ATA/ATAPI controllers / Storage controllers. Look for yellow warning icons. Right-click a device, choose Properties > Driver > Driver Details and note driver date and version. For storage, open Command Prompt as Administrator and run chkdsk C: /scan and wmic diskdrive get status. For NVMe drives, check health with the manufacturer’s tool and confirm firmware is current.
6. Run System File and Image Checks
Open Command Prompt as Administrator. Run sfc /scannow and allow it to complete without closing the window. If it reports unfixable errors, run DISM /Online /Cleanup-Image /RestoreHealth and then run sfc /scannow again. Log files are at C:\Windows\Logs\CBS\CBS.log.
Fixes by Cause
Each fix is structured to be testable and reversible. Change one item, test for stability for at least 30 minutes of mixed idle and load, then proceed.
Unstable RAM and XMP/EXPO Settings
What causes it: The BIOS is applying a memory overclock the CPU memory controller or motherboard cannot stabilize, often because XMP or EXPO pushes frequency, voltage, or timings beyond what your specific silicon handles, especially with four DIMMs.
How to check: Enter BIOS by pressing Del or F2 during boot. Go to AI Tweaker / OC Tweaker / Extreme Tweaker and see if A-XMP, XMP, or EXPO is set to Profile 1 or Profile 2. Compare shown DRAM Voltage to the kit rating on the DIMM label, typically 1.25V to 1.45V for DDR4 and DDR5.
What to do: Set XMP/EXPO to Disabled to run at JEDEC baseline, usually 4800 MT/s for DDR5 or 2133 to 2400 MT/s for DDR4. Save and exit, then test. If stable, re-enable the profile but manually lower frequency one step, for example from 6000 to 5600 MT/s, and set DRAM Voltage to the kit’s rated value exactly. Also set Memory Controller Voltage (VDDIO / SOC Voltage) to Auto initially. For two-DIMM kits, ensure DIMMs are in slots A2 and B2 per motherboard manual.
How to undo: Return to BIOS and re-enable your prior XMP/EXPO Profile or use Load Optimized Defaults under Exit to restore previous auto settings.
Driver Conflict or Bad GPU/Chipset Driver
What causes it: A new driver writes incorrect data to kernel memory, or two drivers compete for the same IRQ or power management routine. GPU and chipset drivers are the most common on custom builds.
How to check: In Device Manager, right-click Display adapters > Properties > Events and check the timestamp versus your crash time. In Settings > Windows Update > Update history > Driver updates, see if Windows installed a driver over your manual install. Use Win + R > verifier to open Driver Verifier Manager only if you are comfortable, as it intentionally stresses drivers.
What to do: Perform a clean driver reinstall. Disconnect from the internet, press Win + X > Device Manager > Display adapters > Uninstall device > Attempt to remove the driver. Then run Display Driver Uninstaller in Safe Mode if needed, selecting GPU > Clean and restart. Reinstall the driver from the official installer using Custom install > Clean installation. For chipset drivers, install the motherboard-specific chipset package from the board vendor under Support > Drivers > Chipset, not a generic package. After install, go to Settings > System > About > Advanced system settings > Hardware > Device Installation Settings and select No to prevent automatic driver overwrites while testing.
How to undo: In Device Manager > Properties > Driver > Roll Back Driver if available, or reinstall the previous driver version you noted before the change. Re-enable automatic driver installs by setting Device Installation Settings back to Yes.
BIOS Incompatibility or Aggressive Auto Overclock
What causes it: An older BIOS lacks microcode for your CPU or memory, or the board applies automatic boosting such as Multi-Core Enhancement, Precision Boost Overdrive, or CPU Lite Load that raises voltage and frequency too far.
How to check: In BIOS, note BIOS Version on the main screen and compare to the CPU support list on the motherboard support page. Check if Precision Boost Overdrive or Multi-Core Enhancement is set to Auto or Enabled.
What to do: Update BIOS only via the board’s recommended method, typically BIOS Flashback or M-Flash / EZ Flash from a FAT32 USB drive with the file at the root. Do not interrupt power. After update, load Optimized Defaults, then set Precision Boost Overdrive to Disabled and CPU Core Ratio to Auto for testing. Set PCIe Mode to Auto rather than forcing Gen 5 if you have older cards or riser cables.
How to undo: Re-enter BIOS and re-enable Precision Boost Overdrive or XMP/EXPO to prior values, or use BIOS > Tool > ASUS CrashFree / MSI M-Flash > Load previous BIOS if your board keeps a backup. Most boards also allow loading a saved profile from OC Profile > Load from USB.
Overheating and Poor Cooling Contact
What causes it: CPU or GPU thermal throttling triggers watchdog timeouts, or VRMs overheat due to insufficient airflow in compact cases. Dried thermal paste or a plastic peel left on the cooler plate also causes sudden spikes.
How to check: In HWiNFO64, watch CPU Tctl/Tdie and GPU Hotspot during a 10-minute Cinebench or in-game load. Check that all fans spin and that pump RPM for an AIO appears under BIOS > Monitor > Fan Info, typically 2000 to 3000 RPM for pumps at full speed. Verify cooler mounting pressure by checking if idle temps drop after reseating.
What to do: Power off and switch off the PSU via the rear 0/1 switch. Clean dust with compressed air, ensure intake and exhaust are not blocked, and confirm front fans are intake and rear/top are exhaust. In BIOS, set fan curves under Monitor > Fan Control > Smart Fan Mode > Manual to keep CPU fan at 60 percent minimum above 70 degrees Celsius. If temps remain high, remove the cooler, clean old paste with isopropyl alcohol, apply a pea-sized dot of new paste, and remount with even cross-pattern tightening.
How to undo: Restore your previous fan curve by selecting Smart Fan Mode > Auto or loading Optimized Defaults. If you replaced thermal paste, no undo is needed beyond restoring fan settings.
Power Supply Instability
What causes it: A PSU that cannot handle transient spikes from modern GPUs, or loose modular cables mixing rails, causes voltage droop and triggers KERNEL_POWER or WHEA errors.
How to check: In Event Viewer > System look for Kernel-Power Event ID 41 without BugCheckCode, which suggests power loss rather than software. Inspect physical connections: the 24-pin ATX, 8-pin EPS12V CPU, and PCIe 12VHPWR or 8-pin PCIe cables should click firmly with no daisy-chained splitters on high-power GPUs unless rated for it. In BIOS Hardware Monitor, watch 12V stability while under load.
What to do: Reseat all power cables at both PSU and component ends. Use separate PCIe cables per GPU power header, not one cable with two pigtails, and ensure the 12VHPWR connector is fully inserted with no visible gap and the sense pins latched. In Windows, go to Control Panel > Hardware and Sound > Power Options > Change plan settings > Change advanced power settings > PCI Express > Link State Power Management and set to Off while testing. Avoid undervolting during diagnosis.
How to undo: Re-enable Link State Power Management to Maximum power savings if you prefer, and return to your original cable routing if you rerouted for testing.
Storage and File System Corruption
What causes it: A failing SSD, loose M.2 standoff, or corrupted NTFS structure causes the kernel to fail when paging data. Overfilled drives with less than 15 percent free space increase this risk.
How to check: Run Command Prompt as Administrator and execute chkdsk C: /scan to scan without reboot, then chkdsk C: /f /r if errors are found, which will schedule a scan on restart. Check Settings > System > Storage > Advanced storage settings > Disks and volumes for drive health warnings. Look at C:\Windows\Logs\CBS\CBS.log after SFC for persistent file errors.
What to do: Free at least 20 percent on the OS drive, update SSD firmware via the vendor toolbox, and reseat the M.2 drive: power off, remove the heatsink screw, lift and reinsert at a 30-degree angle, secure with the standoff, and replace the thermal pad if compressed. If chkdsk repairs fail repeatedly, clone to a new drive before further testing.
How to undo: Firmware updates cannot be undone easily, but you can revert chkdsk changes only by restoring from a prior System Restore point at Control Panel > Recovery > Open System Restore.
When to Replace or Reinstall
Reinstall Windows only after hardware tests pass. If MemTest86 shows errors at JEDEC speeds, if HWiNFO64 reports repeated WHEA errors at stock BIOS defaults, or if chkdsk and vendor tools report reallocated sectors, replace the faulty component first. A reinstall will not fix bad RAM or a failing drive.
Consider a clean install when SFC and DISM cannot repair after two cycles, when Driver Verifier consistently flags the same third-party .sys that clean installs do not fix, or when crashes began immediately after a major Windows Update and Roll Back fails. To reinstall cleanly, back up to an external drive, create installation media with the Windows Media Creation Tool, boot from USB via F8 / F11 Boot Menu, delete old partitions on the OS drive during setup, and install to unallocated space. After install, add drivers one by one in this order: chipset, then GPU, then network and audio, testing stability after each. Keep System Protection enabled at System Properties > System Protection > Configure > Turn on system protection with 5 to 10 percent disk space so you have restore points.
Replace hardware when you have isolated it: RAM that fails solo in every slot, a GPU that crashes in another known-good PC with a clean driver install, a PSU that shows 12V below 11.4V under load, or an SSD with health below vendor warranty threshold. For custom builds under retailer return windows, document your tests with photos of BIOS settings and Event Viewer logs to support an RMA.
FAQ
Should I enable XMP or EXPO on a new custom build?
Enable it only after confirming stability at JEDEC defaults for a day. If XMP or EXPO causes blue screens, treat it as an overclock, not a guarantee. Try the lower of the two profiles, confirm DRAM Voltage matches the kit, and if needed drop frequency one bin. Two DIMMs are more stable than four, and mixing kits, even with the same model number, often fails.
How do I read a minidump without being a developer?
Go to C:\Windows\Minidump, copy the newest .dmp to your desktop, and open it with WinDbg Preview from the Microsoft Store. Use File > Open dump file, then type !analyze -v in the command box and press Enter. Focus on MODULE_NAME and IMAGE_NAME near the bottom. If it points to ntoskrnl.exe alone, the real driver likely corrupted memory earlier and testing RAM and drivers is the next step.
Can undervolting or overclocking cause blue screens even if games run faster?
Yes. An undervolt that is stable in one game can crash in a lighter load when voltage drops too low at idle, and a CPU overclock that passes a short stress test can fail with WHEA_UNCORRECTABLE_ERROR under combined CPU and memory load. Revert to BIOS Optimized Defaults and test. If crashes stop, reapply tuning in small steps, for example CPU offset in 0.01V increments, and test with both heavy and idle periods.
What is the safest way to update drivers on a custom build?
Update one driver at a time and create a restore point first via Create a restore point > Create. Install the motherboard chipset driver from the board vendor, then the GPU driver using Custom > Clean installation, and restart after each. Avoid installing multiple driver utilities that auto-update in the background, such as two different RGB or hardware control suites, because they often install conflicting kernel drivers that trigger SYSTEM_SERVICE_EXCEPTION.
