Inside the lab follow MDrepairs on YouTube, TikTok & Instagram - visit our YouTube channel ↗

Data Recovery Guide

Windows "Critical Process Died" Loop? What It Means for Your Drive

Updated November 18, 2026
In this article

"Critical Process Died" is a Windows blue screen that shows up when a system process the OS needs to stay running has crashed. The stop code is CRITICAL_PROCESS_DIED (0x000000EF). It looks like a software problem. A failing drive can return read errors when Windows tries to load a critical process. Telling those two scenarios apart matters a lot. Software cause: safe to use Windows recovery tools, your data is fine. Hardware cause: stop using the drive, every retry makes things worse.

Before you do anything else, pull your data off the drive first. Once you have your files copied to a separate USB drive (we'll cover how), you can run whatever Windows recovery you want without worrying about losing photos and documents. The big risks are running chkdsk on a failing drive, clicking "Reset this PC" before extracting files, and looping on Startup Repair attempts that keep crashing the same way. All three can permanently lose data that was perfectly recoverable an hour ago.

What "Critical Process Died" Actually Means

Windows has a list of "critical" processes. These are user-mode processes that the kernel depends on to keep the system stable. If one of them dies for any reason, Windows can't continue and throws a CRITICAL_PROCESS_DIED blue screen. Some of the processes that fall into this category:

  • csrss.exe (Client/Server Runtime Subsystem)
  • wininit.exe (Windows Start-Up Application)
  • services.exe (Service Control Manager)
  • lsass.exe (Local Security Authority)
  • smss.exe (Session Manager)
  • winlogon.exe (Windows Logon)

If any of these crashes (or fails to start), Windows commits suicide on purpose with this blue screen. The kernel's reasoning is: I can't safely keep running with one of these dead, so I'll bail out and hope a restart fixes it.

The crash itself doesn't tell you why the process died. The "why" matters a lot. A process can die because of a software bug in the process, a corrupted DLL it depends on, an antivirus driver interfering, a bad Windows update, or because the storage system returned an I/O error while the process was trying to read a file from disk. The first four are software issues. The last one is a hardware issue. Your job is to figure out which group you're in before you try to fix it.

Software Bug or Failing Drive?

Here's the distinction that determines everything else. The same blue screen can come from very different root causes. The tells:

Symptom PatternLikely CauseRisk to Data
BSOD appeared right after a Windows updateUpdate bug or driver mismatchLow. Drive is fine.
BSOD appeared right after installing new software (especially antivirus)Software conflictLow. Drive is fine.
BSOD on every boot but Safe Mode worksDriver or service issueLow. Software fix possible.
BSOD on every boot AND Safe Mode also crashesDeeper system corruption OR failing driveMedium. Investigate drive health.
BSOD at random times during normal useMemory error, driver bug, or drive read errorsMedium. Test memory and drive.
BSOD with different stop codes each timeHardware issue (RAM, drive, or PSU)High. Hardware diagnosis needed.
PC also has been slow to open files latelyDrive is struggling to readHigh. Drive likely failing.
You hear clicking from the drive (HDD)Mechanical failure imminentCritical. Stop using NOW.

The pattern that matters most: does Safe Mode work? Safe Mode loads a minimal set of drivers and services. If the BSOD only happens during a normal boot but Safe Mode is fine, it's almost certainly software. If Safe Mode also crashes, something deeper is wrong, and that's when you have to take the drive failure possibility seriously.

Is Your Data at Risk?

If your CRITICAL_PROCESS_DIED is software-caused: your data is fine. Photos, documents, downloads, all sitting safe on the drive. You can fix Windows however you want and the user files come along for the ride.

If your CRITICAL_PROCESS_DIED is hardware-caused (the drive is failing): your data is at active risk every time you power the PC on. Each boot attempt that crashes is the drive trying and failing to read system files. That same wear can damage your personal files too. The longer you let the PC keep trying to boot, the worse the drive gets.

Cases where data definitely IS at risk regardless of cause:

  • You ran chkdsk /f /r and the drive had bad sectors. Chkdsk on a failing drive can scramble the file system.
  • You clicked "Reset this PC" or "Cloud download" and let it run. Both wipe C:.
  • You let Windows do a "Repair install" and selected the option that doesn't keep files.
  • You forced 10+ power cycles trying to get past the BSOD. Each forced power-off on Windows mid-write can damage file system structures.

What NOT to Do Right Now

Do NOT:
  • Run chkdsk /f or chkdsk /r without verifying drive health first. Chkdsk on a sick drive can damage the file system metadata. Diagnose drive health before running chkdsk.
  • Click "Reset this PC" in the recovery menu. Both "Keep my files" and "Remove everything" can fail and lose data. "Cloud download" is the worst option for data preservation.
  • Run Startup Repair more than twice. Startup Repair writes to the drive each attempt. On a failing drive, repeated writes can finish the drive off.
  • Boot a Windows install USB and click "Install Windows". Choosing a partition to install onto and formatting it wipes your data.
  • Use sfc /scannow or DISM /RestoreHealth on a drive you suspect is failing. Both write data. Both can fail destructively if the drive is sick.
  • Force-shutdown the PC 5+ times to "trigger Automatic Repair". This is an internet myth that wastes drive life. Each forced shutdown stresses the drive.
  • Disable the storage driver in Safe Mode. If you can't boot the drive afterward, you have no way to access the data.
  • Update the BIOS or firmware during this crisis. A failed update on a struggling PC turns it into a brick.

Test 1: Try to Boot Into Safe Mode

This is the first diagnostic. If Safe Mode boots, your drive is likely healthy and the issue is software. If Safe Mode also crashes with the same BSOD, that's a warning that something deeper is wrong.

Step 1: Get to the Windows Recovery Environment

If the PC is stuck in a BSOD loop, after three failed boots Windows usually launches Automatic Repair on its own. From there click "Advanced options". If you can boot normally to the lock screen but get the BSOD shortly after, hold Shift and click Restart from the start menu.

Step 2: Navigate to Startup Settings

In the Recovery Environment, click Troubleshoot, then Advanced options, then Startup Settings, then Restart. The PC reboots into a menu of numbered boot options.

Step 3: Pick Safe Mode

Press 4 (or F4) for plain Safe Mode. Press 5 (or F5) for Safe Mode with Networking if you need internet access in Safe Mode to download tools. Wait for boot.

If Safe Mode boots and you reach a desktop: drive is likely fine, problem is in the regular boot drivers or a recently installed app. Run the software fixes below.

If Safe Mode crashes with the same CRITICAL_PROCESS_DIED: drive may be failing. Skip the software fixes for now. Do the Linux Live USB diagnosis next.

Test 2: Linux Live USB Drive Health Diagnosis

A Linux Live USB lets you bypass Windows entirely and ask the drive directly: are you healthy? You boot a small Linux operating system from a USB drive, read the SMART data, and look for warning signs. The drive doesn't get modified at all. This is the safest possible diagnostic.

Step 1: Build a Linux Live USB

On a working computer, download Ubuntu 24.04 LTS from ubuntu.com/download/desktop. Use balenaEtcher or Rufus to write the ISO to an 8GB USB drive. This creates a USB that boots a full operating system without installing anything on the broken PC.

Step 2: Boot the Broken PC from the USB

Plug in the USB. Power on. Hit the boot menu key (F12 on Dell/Lenovo, F11 on HP, Esc on Acer). Pick the USB. Choose "Try Ubuntu" not "Install Ubuntu".

Step 3: Read the Drive's SMART Data

Open a terminal (Ctrl+Alt+T). Type `lsblk` to see drives. Your internal SSD is usually `/dev/nvme0n1` and SATA SSDs/HDDs are `/dev/sda`. Run `sudo smartctl -a /dev/nvme0n1` (or substitute the right device). Look at the output for: Reallocated_Sector_Ct, Current_Pending_Sector, Offline_Uncorrectable, and on NVMe drives, critical_warning and media_errors.

Step 4: Interpret the Results

All those fields at 0: drive is healthy. The BSOD is software. Move on to software fixes. Anything above 0: drive is failing. Skip software fixes. Extract files immediately (next section) and replace the drive after.

If smartctl shows the drive is failing: Power the PC off as soon as you've finished imaging/copying your data. Every additional power-on cycle accelerates the failure. Order a replacement drive and plan to image the failing one to the new one as soon as possible. Our BSOD-from-failing-drive guide covers more on this scenario.

Extract Your Files Before You Do Anything Else

Whether the cause is software or hardware, this step is the same. Copy your files off the drive while you still can. Then you have options. Without files copied, every fix attempt is a gamble.

You're already booted into Ubuntu from the previous step (or you can boot back into it). Now you copy files.

Step 1: Plug in an External USB Drive

Use any USB hard drive or large USB stick with enough space to hold your important data. Ubuntu will mount it automatically in the Files application.

Step 2: Find the Windows Partition

In Files (the file manager on Ubuntu's left sidebar), look for a partition labeled "Windows" or that contains folders like "Program Files", "Users", "Windows". Click to mount it. Browse into Users, then into your username's folder.

Step 3: Copy the Important Folders

Select Documents, Desktop, Downloads, Pictures, Videos, Music. Drag them to the external drive in Files. If you have specific app data (browser bookmarks, email, OneDrive offline cache), press Ctrl+H to show hidden folders and copy AppData/Local and AppData/Roaming too. The full AppData copy is large but worth it.

Step 4: Skip Bad Files Without Aborting

If a specific file errors out during copy, click "Skip" not "Cancel". One bad file should not stop the rescue. If many files error out, the drive is failing badly and you should stop and send it to a lab for controlled imaging.

Step 5: Verify on a Working PC

Unplug the external drive. Plug it into a working PC. Open it. Open a few files to confirm they work. Only then proceed to fix the broken PC.

When chkdsk Helps and When It Hurts

chkdsk has a reputation as the universal Windows fix. It's not. On a healthy drive with file system errors, it's helpful. On a failing drive, it's destructive.

When chkdsk Helps

Your drive is healthy (SMART data clean). You've had a power outage during file operations, a forced shutdown, or a crash that left the NTFS in an inconsistent state. chkdsk can correct the inconsistencies and bring the file system back to a clean state. No data should be lost in this case.

When chkdsk Hurts

Your drive has bad sectors. chkdsk's job is to find and mark bad sectors, but in the process of moving data away from them, it can damage adjacent metadata. On a drive that's already failing, chkdsk frequently does more harm than good. The most common terrible outcome: chkdsk runs for hours, finishes, and you reboot to find that your Documents folder is now empty (the directory entries were corrupted during the operation).

The /r flag is especially dangerous on failing drives. /r tries to recover information from bad sectors. On a healthy drive that's mostly fine, this works. On a failing drive where many sectors are bad, /r can run for days and damage the file system structure beyond recovery.

Safe rule: Don't run chkdsk on a drive you haven't verified is healthy. Run smartctl from Linux first. If SMART is clean, chkdsk /f is fine. If SMART shows trouble, send the drive to a lab.

Software Fixes (Once Data Is Safe)

Now that you've copied your files off, you can try to fix Windows. If anything goes wrong, your data is safe somewhere else. Here are the software-only fixes in rough order of safety.

Uninstall the Most Recent Update

If the BSOD started right after a Windows update, uninstall the update. From the recovery environment, go to Troubleshoot, Advanced options, Uninstall Updates. Choose either the latest quality update (cumulative patches) or the latest feature update. Pick quality first if the BSOD only appeared this week. Feature update if it's been months.

Boot With Last Known Good Configuration

If you have System Restore points, use one from before the problem started. In the recovery environment, click Troubleshoot, Advanced options, System Restore. Pick a date that's before the BSOD started showing up. System Restore doesn't touch user files, only system files and the registry.

Run sfc and DISM

If you've verified the drive is healthy and Safe Mode boots, open an admin command prompt in Safe Mode and run `sfc /scannow`. This scans Windows system files and replaces corrupted ones. After it finishes, run `DISM /Online /Cleanup-Image /RestoreHealth`. This downloads fresh copies of Windows system files from Microsoft's servers (if the PC has internet in Safe Mode) and repairs the system image. Both of these can fix software corruption that causes CRITICAL_PROCESS_DIED without losing data.

Disable Recent Drivers

If the BSOD started after a driver install, boot to Safe Mode, go to Device Manager, find the recently-installed driver, right-click, Properties, Driver tab, Roll Back Driver. Or just uninstall it. Network adapters, GPU drivers, and storage controller drivers are the most common culprits.

Reset Without Removing Files (Last Resort, AFTER You've Copied Files Off)

If nothing else works and you've already copied your data to an external drive, you can use "Reset this PC" with "Keep my files" as a final attempt. It reinstalls Windows but tries to preserve user files. The reason we say AFTER copying files off is that this option sometimes fails midway and leaves the drive in worse shape. With files already backed up, the worst case is just doing a fresh install.

Inside the Lab: When the Cause Is Actually Hardware

If diagnosis shows the drive is failing, here's what happens when you send it to MDrepairs.

Intake and Power-On Analysis

We connect the drive to a hardware imager that monitors current draw and the drive's behavior before any read attempt. A failing drive often reveals itself in the power-on stage. Spinning hard drives that click are obvious. SSDs that draw weird current at idle are diagnosable from the rig. We diagnose before we read.

Controlled Imaging

We take a sector-level copy of the drive to a separate disk in the lab. For sick drives, we use slow-read mode that gives the drive time on each sector. We skip bad sectors after limited retry attempts instead of letting them hang and stress the drive further. The original drive gets put aside once imaging is done. All recovery work happens on the image.

Reading SSD NAND Directly When Needed

Modern SSD recovery depends on a viable controller and encryption path. The lab assesses original-board repair, controlled reading, and firmware or virtual-translator access supported for the exact device. Independent raw-NAND chip-off cannot bypass encryption, and a lost controller path can make recovery impossible.

File System Reconstruction

Once we have an image, we rebuild damaged NTFS metadata using backup copies that NTFS maintains. Then we walk the directory tree and extract files. For BitLocker-encrypted drives, we apply your recovery key to decrypt before extraction.

Returning the Data

We deliver recovered data on an external drive (yours or one we supply). Files are organized as they were on the original drive. You can drag the recovered folders onto a new PC and resume work.

How Much Does Critical Process Died Recovery Cost?

ScenarioTypical CostWhy
Software cause, you fixed it yourself$0 to $50USB drive + external for backup
Software cause, you brought it to a shop for the Windows fix$100 to $300Local PC repair shop work, not lab recovery
Lab logical recovery (filesystem damaged from failed chkdsk)$1,000 to $1,500Imaging plus NTFS reconstruction
Failing SSD, controller alive, NAND alive$1,000 to $1,500Slow imaging with read-error handling
Failing HDD with bad sectors and slow reads$1,000 to $3,000+Slow imaging, possible head clean room work
Original-board / supported controller recovery$1,500 to $2,000+Requires a viable original reading/encryption path; chip-off is not a general fallback
HDD with mechanical failure (head crash, seized motor)$1,500 to $3,000+Clean room work, donor parts

The $100 diagnostic is credited toward a successful recovery and is non-refundable once diagnostics start. Approved donor or parts charges are non-refundable once ordered, even if recovery fails. You approve the estimate before recovery work starts. Free phone consult. Full pricing across scenarios.

Stop This From Happening Again

Image-Based Backups, Once a Week Minimum

Macrium Reflect Free, Acronis True Image, or Veeam Agent. Image-based means the backup is a snapshot of the entire drive: Windows, programs, user files, all of it. With an image-based backup, the CRITICAL_PROCESS_DIED loop is a 20-minute restore instead of a multi-day ordeal. Schedule it weekly to an external drive or NAS.

Don't Install New Antivirus or System-Level Software Right After a Windows Update

Let the new Windows version settle for a few days before adding new third-party drivers or kernel-level software. Antivirus, VPN clients, and disk utilities install kernel-mode drivers that can conflict with updated Windows internals. Stagger your changes.

Check Drive Health Quarterly

Install CrystalDiskInfo (free) and run it once a quarter. It reads SMART data and gives you a simple Good / Caution / Bad rating. When it switches to Caution, plan a replacement before it hits Bad.

Don't Run Windows on a Drive That's More Than 90% Full

Windows needs free space for system file updates, temp files, and crash dumps. A drive that's at 99% can throw all kinds of weird errors including CRITICAL_PROCESS_DIED if a system process can't write to a temp file it needs.

Replace Aging Hard Drives Before They Die

Spinning HDDs have a finite mechanical life. Most last 3 to 7 years in daily use. Past year 5, the failure risk increases sharply. If you're still running a hard drive from before 2020, plan to replace it (or at least back it up to a newer drive) this year. SSDs last longer but also have endurance limits, measured in TBW (terabytes written).

FAQ

Is "Critical Process Died" usually a hardware problem or a software problem?

Software and drive faults are both possibilities. The Safe Mode test is the fastest way to tell. If Safe Mode boots, software. If Safe Mode crashes too, investigate hardware.

Will sfc /scannow fix Critical Process Died?

Sometimes. sfc replaces corrupted Windows system files with known-good copies. If the BSOD is from a corrupted DLL or system file, sfc can fix it. It can't fix hardware issues. And running sfc on a failing drive can make things worse because it writes data.

Can I just install Windows fresh on a new drive and copy files from the old drive?

Yes, this is often the cleanest approach when the old drive is failing but still readable. Install fresh Windows on a new drive, then connect the old drive via USB enclosure or as a second internal drive, and copy what you need. This is faster than recovery in most cases.

Why does the BSOD only happen sometimes?

Intermittent BSODs are usually hardware. Failing drives don't fail consistently. Some sectors are fine, some are slow, some return errors. When a "Critical Process Died" only happens randomly, that pattern points at hardware. Test the drive.

I see "Critical Process Died" and a different stop code on different boots. Bad sign?

Yes. Different stop codes between attempts suggest the underlying problem is hardware (RAM, drive, PSU). Software issues tend to be repeatable. Test memory with the built-in Windows Memory Diagnostic and test the drive with SMART data.

Do I need to send the whole laptop or can I just send the drive?

Just the drive is enough in most cases. Pull it out, pack it well, ship it. We image and recover the data, ship the recovered data back to you. How to ship for recovery walks through it. If the drive is soldered to the motherboard (some thin laptops), then you ship the whole laptop.

What if the drive is BitLocker-encrypted?

You need the BitLocker recovery key. It's usually saved to your Microsoft account at account.microsoft.com/devices/recoverykey. Without it, encrypted drives are mathematically unrecoverable. We don't accept BitLocker cases without the key because we can't recover them.

My PC was working when I shut it down. I powered it on the next day and got this BSOD. What happened?

Most likely a Windows update finished installing during shutdown and the next boot is the first time the new files are loading. Sometimes there's a bug in the update or in the way it interacted with your specific hardware. Try uninstalling the most recent update from recovery options.

Should I run the Windows Memory Diagnostic too?

Yes if you suspect memory problems. Type "Windows Memory Diagnostic" in the start menu, click "Restart now and check for problems", let it run. If it finds errors, your RAM may be the cause and replacement is needed.

How long does the Linux Live USB diagnosis take?

About 90 minutes if you've never done it. Download Ubuntu (4GB), write to USB (10 minutes), boot the PC (3 minutes), run smartctl (instant), copy files (varies by size). The slow parts are the download and the file copy.

Do you handle mail-in cases from outside New Jersey?

Yes. Free UPS 2nd Day Air shipping (contiguous US) — both ways, insured. Similar laptop won't boot scenarios are here.

MDrepairs is an independent data recovery lab in New Jersey serving customers nationwide via mail-in. We're not affiliated with Microsoft. We get data off broken Windows computers, even when the local repair shop says it's a lost cause.

Lost your data?
Let's assess your options.

Estimated turnaround: 4–6 weeks; timing may vary by case. Start with a free initial quote. The $100 lab diagnostic is credited on success; diagnostic and approved parts charges apply on failure.

Split the cost of your recovery with Affirm. Explore financing Subject to eligibility.

Devices recovered at our lab

These rows come from BenchKey, the ticket system we built for our own shop. Customer names, contact details and descriptions are left out.

Phone logic board with three removed chips at our Lincroft lab
A phone logic board sits behind three removed chips, solder pads up, at our Lincroft lab.
4.6 · 345 Google reviews$100 diagnostic, credited toward a successful recovery
Start your mail-in recovery