My hard drive started making strange noises and now it shows bad sectors, and I’m worried I could lose important files. I need help figuring out the safest way to recover data from a failing hard drive before it gets worse. Any advice on data recovery steps, software, or whether I should stop using the drive would really help.
CRC errors and folders freezing, what I did
I ran into this more than once. You click a folder, Explorer locks up, or Windows throws a cyclic redundancy check error and your stomach drops. I know the feeling. When I saw it the first time, I kept retrying out of panic. Bad move.
The first thing I do now is stop touching the drive.
If it is your system drive, power the machine off. If it is an external drive, unplug it. Leaving it online gives the drive and the OS more chances to hammer weak sectors. I saw one drive go from ‘opens slowly’ to ‘not readable at all’ after a long afternoon of retries, previews, and background indexing. I still regret that one.
What those bad sectors usually mean
There are two buckets here.
Soft bad sectors are usually data consistency problems. The sector content no longer matches its error correction data. I have seen this after a power cut and after someone held the power button during a write. Sometimes those recover cleanly.
Hard bad sectors are physical trouble. Platter damage, weak head, head crash starting, aging media, stuff like that. Software does not repair physical damage. At best, the drive gets told to avoid those spots later.
That difference matters because it changes your goal. Your goal is not to heal the disk first. Your goal is to pull data off before the situation gets worse.
My rule now, clone first
I would not start with CHKDSK. I know people reach for it fast because it is built in and familiar. I did too. It can make the file system mountable again, sure, but it does not care about preserving every file entry the way you do. If metadata is damaged, it may throw things away to make the structure consistent.
So I treat the source drive like evidence. Read from it as little as possible. Make one full copy attempt onto a healthy disk, then work from the copy.
A byte for byte clone or image is the safer path. If the original drive gives up halfway through your later tests, at least you are not starting from zero.
What worked better for me
For DIY recovery, I had better luck with tools built for imaging damaged drives than with repair tools. One option is Disk Drill. A lot of people know it for deleted files, but the part I would look at here is the Drive Backup feature.
What stood out to me in version 6 was how it handles unstable areas. When it hits a bad block, it does not sit there forever and stall the whole job. It skips rough spots, grabs the readable areas first, then circles back and retries those trouble zones with smaller chunks. On a sick drive, that matters. Less strain, more data pulled before the thing taps out.
The short DIY path
1. Get another drive ready
Use a healthy destination drive with at least the same capacity as the failing one. More is fine. Less is a dead end.
2. Create an image or clone
Start the backup and wait. No multitasking on the source if you can avoid it. Imaging a damaged disk is slow. I have seen it take hours. I have also seen it drag into the next day.
3. Scan the image, not the original
Once you get an image file like .dmg or .img, mount or attach it in your recovery software and scan that file. This keeps wear off the original disk.
4. Recover to a different place
Save recovered files to a third drive or location. Not back to the failing drive. Not onto the clone source target if you can help it. Keep the recovered output separate so you do not overwrite anything useful.
When I would stop and send it out
If the drive clicks, grinds, beeps, or does not spin up right, I would quit the home recovery attempt. Same if BIOS stops seeing it consistently. Those are not good signs.
At that point, software is not the tool. A lab is.
Places like Gillware or Techchef handle mechanical failures in cleanroom conditions. They swap heads, move platters, deal with controller issues, all the ugly stuff most of us should not touch. The bill is often around $500 to $3,000 from what I have seen. Painful, yes. Still cheaper than losing business records, family photos, or project data you cannot rebuild.
After the files are safe
Only after I verified my files would I try any repair step on the original drive.
If you still want to experiment with it, Windows gives you a couple of options:
chkdsk /rin PowerShell or Command Prompt, which checks for bad sectors and marks them- A full format, not a quick one, which forces Windows to read through the disk and mark damaged areas as unusable
I would not trust that drive again for anything important. Once bad sectors start showing up, I treat the drive like it already gave notice.
What I changed after getting burned
I stopped pretending one copy was enough.
Use a 3-2-1 backup setup:
- 3 copies of your data
- 2 different storage types
- 1 copy off-site
That sounds boring until a drive starts freezing folders and throwing CRC errors. Then it sounds cheap.
8 Likes
Stop using the drive. Strange noises change the math. Bad sectors alone are one thing. Noise points to hardware failure, and every extra spin-up hurts your odds.
I agree with @mikeappsreviewer on avoiding CHKDSK first. I disagree a bit on one point though. If the files are irreplaceable and the drive is clicking or grinding, I would skip DIY sooner. Labs cost a lot, but repeated home attempts often make the final recovery bill worse.
What I’d do:
-
Check SMART once, fast.
Use CrystalDiskInfo or smartctl. If you see reallocated sectors, pending sectors, or read error counts climbing, stop testing. One quick look, then move on. -
Keep the drive cool and stable.
Use a desktop SATA port if possible, not a flaky USB hub. Put the drive on a flat surface with airflow. Heat kills weak drives faster. -
Copy the most important files first.
Before a full image, I’d try a triage pass only if the drive is still readable and not clicking nonstop. Grab docs, photos, databases, stuff you cant replace. Big video files come later. -
Then image it.
Disk Drill is fine for this because it can work around unreadable sectors and let you scan the image instead of the dying disk. That part matters more than the file recovery branding. -
Verify recovered files.
Open samples. Check file sizes. CRC errors love to leave you with junk files that look normal untill you open them.
For context, what hard drive bad sectors and unreadable blocks mean is worth reading.
Do not freeze the drive. Do not run defrag. Do not keep rebooting to “see if it works now.” Thats how drives die on your schedule, not yours.
Strange noises changes this from ‘maybe recoverable at home’ to ‘be careful right now.’
I mostly agree with @mikeappsreviewer and @viaggiatoresolare, but I’d push one thing harder: if the drive is clicking, scraping, or spinning down randomly, stop trying to browse folders entirely. Even ‘just checking what’s still there’ can make it worse. People underestimate how much Windows loves poking a sick drive in the background.
A couple things I’d add that they didn’t really stress:
- Disable auto-mount/indexing stuff if you reconnect it
- If it’s an external HDD, try a different cable and power supply first, because sometimes ‘bad sectors’ gets mixed with flaky USB communication
- If the data is super high value, skip the full DIY experiment and go straight to a lab before the heads get uglier
I actually disagree a bit with the ‘copy important files first’ idea unless the drive is still fairly calm. On a really unstable disk, cherry-picking files causes extra seeking all over the platter. A sequential image is often kinder to the hardware than hopping folder to folder. That’s why imaging tools matter more than normal file copy.
If you want the safest home route, use Disk Drill to make a byte-level backup of the failing drive, then scan the image instead of the original disk. That’s the main benefit here, not magic repair. Data recovery from hard drive bad sectors is mostly about reducing stress on the dying drive, not ‘fixing’ it.
Also do a quick SMART health check before you decide how far to push it. This explains why SMART can say a drive looks fine while your data says otherwise: why checking SMART status still matters before hard drive recovery
Stuff I would not do:
- CHKDSK first
- Defrag
- Full format
- Repeated reboots
- Freezer nonsense from 2006 YouTube lol
If the drive disappears from BIOS, starts loud clicking, or takes forever to identify, that’s lab time. At that point software is mostly wishful thinkng.
I’m with @viaggiatoresolare and @cacadordeestrelas on one big point: noise means stop gambling. But I’d push back a little on doing even a quick file triage first if the drive is actively clicking. Random file copying makes the heads jump around more. A straight image attempt is usually the gentler move.
My take:
- If it’s grinding, clicking, or dropping out of BIOS, skip DIY
- If it’s readable and only lightly degraded, image first, recover second
- If it’s an external drive, rule out the enclosure before blaming the disk itself
That last one gets missed a lot. Pull the drive from the USB enclosure only if you know what you’re doing and warranty does not matter. Bad USB bridge boards, weak power bricks, and sketchy cables can mimic disk failure symptoms.
One thing I would add that @mikeappsreviewer did not really stress: avoid mounting it in your normal everyday OS session if possible. Boot from another drive or a live USB so the sick disk is not getting indexed, thumbnailed, scanned by antivirus, or casually poked by Explorer.
If you do try software, use something that can make an image and then work from the image. Disk Drill fits that use case.
Pros of Disk Drill:
- easy imaging workflow
- can scan the image instead of the original drive
- decent for pulling files from damaged file systems
Cons:
- not a substitute for hardware repair
- deep scans can take forever
- on seriously failing drives, dedicated cloning tools may still be better
So yes, Disk Drill is reasonable here, but only while the drive is still readable. Once the hardware side gets uglier, software stops being the answer.