CRITICAL_PROCESS_DIED Blue Screen Fix (Win 10/11)

You're mid-render in Premiere, or you just clicked 'Restart' after a Windows Update, and the screen goes blue with a sad face and the words CRITICAL_PROCESS_DIED. The system reboots before you can read the stop code. It happens again. And again. Sometimes it loops straight back to the same blue screen before you even reach the login prompt.

This isn't the kind of BSOD that pops up once and vanishes. It usually appears right after a driver install (GPU drivers are the usual suspect), a Windows cumulative update, a new stick of RAM, or a corrupted system file that's been lurking for months. The trigger matters, because the fix depends on it.

What's actually happening here

Windows has a handful of processes that are non-negotiable. csrss.exe, wininit.exe, smss.exe, services.exe, lsass.exe. These aren't user apps — they're the skeleton the rest of the OS hangs off. If any of them terminates unexpectedly, Windows can't keep running safely, so it deliberately bugchecks. That's the blue screen. It's not a crash in the normal sense; it's Windows pulling the plug on itself because continuing would be worse.

The reason step 3 works below is that it forces Windows to rebuild the exact files these processes load from. If the underlying binary or its dependencies are corrupted, patching the registry or updating a driver won't help — you need to replace the file itself.

If you see this BSOD on every boot, not just occasionally, skip straight to Step 4. The other steps assume you can reach the desktop.

Fix it in order

Boot into Safe Mode. Hold Shift while clicking Restart, then choose Troubleshoot → Advanced Options → Startup Settings → Restart → press 4. Safe Mode loads a stripped-down driver set. If the BSOD disappears here, you've narrowed it to a third-party driver or startup app. That's useful information.
Run SFC and DISM. Open an elevated Command Prompt (right-click → Run as administrator) and run these one at a time, waiting for each to finish:
```
sfc /scannow
dism /online /cleanup-image /restorehealth
sfc /scannow
```
The second sfc run matters. DISM repairs the component store that SFC pulls replacement files from. Running SFC first against a broken store sometimes fixes nothing. The order in that block is deliberate.
Check your disk. A dying SSD or a mechanical drive with reallocated sectors will hand Windows corrupted binaries and cause this exact bugcheck. Run:
```
chkdsk C: /f /r
```
It'll ask to schedule on next reboot — say yes. On a 1TB drive this can take an hour. If chkdsk reports bad sectors, back up now and replace the drive. I've seen three of these BSODs in the last year that were nothing but a failing Samsung 970 EVO.
Roll back recent changes. If this started after a driver update, open Device Manager, find the device (GPU, network adapter, chipset), right-click → Properties → Driver tab → Roll Back Driver. If the button is greyed out, uninstall the device and reboot — Windows will reinstall the inbox driver.

For a Windows Update that arrived in the last 10 days, go to Settings → Windows Update → Update history → Uninstall updates. A specific cumulative update is often the culprit; you can pause updates for up to 5 weeks while you wait for the follow-up patch.
Test your RAM. Faulty memory corrupts arbitrary pages in kernel space. Press Win+R, type mdsched.exe, choose 'Restart now and check for problems'. Or better, if you have a spare USB stick, grab MemTest86 and run four full passes. Windows Memory Diagnostic is fine for catching obvious failures but misses marginal DIMMs that only fail under specific timings.

If you recently enabled XMP or DOCP in BIOS, disable it and test again. Instability from aggressive memory profiles causes this BSOD more often than people admit.
Reset the OS as a last resort. Settings → System → Recovery → Reset this PC → Keep my files. This reinstalls Windows while preserving your documents. If it still BSODs after reset, the hardware is the problem, not the software.

If you can't even reach Safe Mode

You need a Windows install USB. Create one on another machine with the Media Creation Tool. Boot from it, choose Repair your computer → Troubleshoot → Advanced Options. From there you get a command prompt pointed at the offline Windows install. You can run:

sfc /scannow /offbootdir=C:\ /offwindir=C:\Windows
dism /image:C:\ /cleanup-image /restorehealth

Adjust drive letters if Windows isn't on C: in that recovery environment — it often isn't. Use diskpart → list volume if you're unsure.

What to check if it still fails

Pull the actual crash dump. It's at C:\Windows\Minidump — you'll need to enable 'Show hidden files' and take ownership of the folder if it's locked. Load the .dmp in WinDbg Preview (free from the Microsoft Store) and run !analyze -v. The MODULE_NAME line near the top tells you which driver or process crashed. That's the single most useful piece of information you can get, and it takes 90 seconds.

If the dump points at ntfs.sys, replace the drive. If it points at nvlddmkm.sys, it's the NVIDIA driver — DDU (Display Driver Uninstaller) in Safe Mode, then a clean install. If it points at ntoskrnl.exe with no clear module, you're usually looking at RAM or a chipset driver problem. Test the RAM first, then update the chipset driver from your motherboard vendor, not Windows Update.

One more thing: check event viewer under Windows Logs → System for Event ID 41 (Kernel-Power). If that shows up alongside your BSODs, you may have a failing PSU, not a software issue at all.