Your drive is acting strange, you search for help, and within two results somebody tells you to open an admin command prompt and run chkdsk /f /r. It’s the most repeated piece of storage advice on the internet.
It is also, in my experience, the single most common way a drive that could have been recovered completely comes to me in pieces.
What CHKDSK actually does
CHKDSK is a file system repair tool. That word — repair — is doing a lot of work. It doesn’t recover your files. It makes the file system internally consistent again, and it does that by discarding anything it can’t reconcile.
When CHKDSK finds a directory record pointing at a file it can’t verify, it doesn’t preserve the relationship. It severs it. Orphaned data gets dumped into a found.000 folder as numbered .chk files with no names, no extensions, and no folder structure. Thousands of intact family photos can come out the other side as FILE0001.CHK through FILE4000.CHK.
The files are technically still there. Everything that told you which one was your daughter’s graduation is gone.
The /r flag is worse
/f fixes file system errors. /r additionally scans every sector on the disk looking for bad ones, and attempts to read each one repeatedly before marking it bad.
On a healthy drive, that’s fine — it’s just slow.
On a drive that’s physically degrading, /r is close to the worst thing you can do. It forces a full-surface read on hardware that is already failing to read. Every retry on a weak sector is more time the heads spend hammering the same struggling area. A drive with a few hundred bad sectors can pick up thousands during a /r pass, and a drive with marginal heads can fail outright partway through — now with a half-modified file system on top of the original problem.
When CHKDSK is fine
I’m not saying never. CHKDSK is a reasonable tool when all of these are true:
- The drive is mechanically healthy — no clicking, no grinding, no slow reads, clean S.M.A.R.T. data.
- The data on it is backed up somewhere else, or you don’t care about it.
- The problem is a minor inconsistency after an unclean shutdown.
That’s the case it was built for: a working disk with a small bookkeeping error. Run it on your boot SSD after a power cut and you’ll almost certainly be fine.
Free evaluation. Firm quote before any work begins.
What to do instead
The correct order of operations on a drive whose contents matter is always the same: image first, repair the copy.
A sector-by-sector image gets the data off the failing hardware and onto something stable, handling bad sectors carefully instead of grinding on them. Then every repair attempt — CHKDSK, file system rebuilds, anything — happens against that copy. If a repair goes wrong, you make another copy and try something else. On the original, there are no second attempts.
This is why a RAW drive or a drive throwing CRC errors should be powered down rather than repaired in place. The repair isn’t the risky part. Doing it to your only copy is.
The bottom line
CHKDSK is a maintenance tool that people reach for as a recovery tool. It fixes the file system, not your files, and its method for fixing is to throw away what confuses it. If the data on the drive matters and you don’t have another copy of it, don’t run it — and definitely don’t run /r on a drive that’s making noise or reading slowly.
Already ran CHKDSK and ended up with a found.000 folder? Start a case — a lot of that is still reconstructible, and it helps enormously if you tell me exactly which flags you used.