HDD bad sector repair: recover data or replace the drive?

“I ran chkdsk and it found bad sectors. Can they be fixed?” This question lands in the shop all the time. Short answer: sometimes yes, sometimes no—and honestly, that’s the less important question. The more important one is what you do right now, before you try to “fix” anything.

Because once a drive starts reporting bad sectors, you’ve got a bigger priority than repair: backup. And no, that’s not just a cliché—you’ll see the numbers in a moment that prove why it matters twice as much.

Let’s break it down the way I explain it to customers in the shop: what a bad sector actually is, what you can and can’t fix, how to spot a dying drive from the numbers—and when the cheapest solution is simply replacing the drive and focusing on the data.

Before you run anything: if the drive holds data you don’t want to lose, back it up now—before you launch another scan. Every surface test puts more stress on a dying drive. When nothing can be read anymore, that’s not a job for repeated scanning; it’s a job for data recovery.

What a bad sector actually is—and why the type matters

A drive is divided into tiny “compartments”—sectors. A bad sector is one the drive can’t reliably read back what it wrote. The catch is that bad sectors come in two flavors, and each means something completely different (Wikipedia sums it up nicely):

• Logical (soft) bad sector. The “hardware” isn’t broken; the data just doesn’t match its checksum. Typically happens when power cuts out during a write—the drive didn’t finish writing and the sector was left in an inconsistent state. This one can be fixed—by overwriting it. The Windows chkdsk tool handles that.
• Physical (hard) bad sector. Actual damage to the platter surface—a scratch, a head crash, degradation of the magnetic layer, wear from age. This one can’t be fixed. It can only be “mapped out”—marked so nothing gets written there anymore.

How do you tell them apart? Honestly, often by trying to fix them. What gets fixed was logical. What remains is physical damage—and that’s a sign the drive is aging.

SMART: three numbers that decide

Modern drives keep internal health statistics—SMART. You can read it for free with a tool like CrystalDiskInfo. Out of dozens of values, three matter most for bad sectors, and together they tell the story of a dying sector:

1. Current Pending Sector Count (197) — “pending” unstable sectors. The drive can’t reliably read them yet and is waiting to decide what to do.
2. Reallocated Sector Count (05) — sectors the drive has judged as bad and remapped to spare ones. The drive’s firmware does this on its own.
3. Uncorrectable Sector Count (198) — sectors from which data couldn’t be recovered. Here you’ve already lost data, whether you noticed or not.

SMART diagnostics show drive health along with the count of reallocated, pending, and uncorrectable sectors. A few dozen may mean nothing—what matters is whether the number grows over time.

And now the most important rule: it’s not the absolute number that decides, it’s the trend. A drive with a few reallocated sectors that hasn’t budged for years can happily keep serving. A drive where the number climbs week by week is on its way out—even if it still “works” for now.

Why SMART isn’t enough (and why backup is king)

Here’s that promised number. Cloud company Backblaze runs hundreds of thousands of drives and has published failure statistics for years. Their analysis of SMART values states clearly: 76.7% of drives gave a warning before failure through one of the key SMART attributes (including reallocated and pending sectors). Sounds good—until you do the rest of the math.

23.3% of drives failed with no warning at all. Nearly one in four died with a clean SMART. That’s why you can’t rely on any diagnostic tool as a guarantee, and why the only real insurance is a backup—ideally following the 3-2-1 rule. SMART is an excellent servant but a terrible master.

Can you “fix” bad sectors on an HDD? What tools actually do

Now for the promised myth. People search for “bad sector repair” and expect some program to iron out the damaged surface. It can’t. Let’s walk through what the tools actually do.

And now the most common—and often most harmful—advice from the internet: chkdsk. Yes, chkdsk C: /f fixes logical file system errors, and chkdsk C: /r scans the entire surface. On a healthy drive after an unclean shutdown, that’s fine. But on a drive that’s already throwing bad sectors, chkdsk can do serious damage—and we see it regularly in the shop:

• A dying drive comes in. chkdsk /r reads the entire drive sector by sector (on a large HDD, that can take hours). Data recovery specialists warn that this extra load often finishes off an already weak mechanical drive—“it reads slowly” suddenly becomes “it won’t spin up at all.”
• It destroys data that could have been saved. chkdsk /f rewrites the file system and “cleans up.” On a damaged drive, it scrambles the structure, dumps unreadable files into a FOUND.000 folder as useless .chk fragments, and makes professional data recovery much harder or outright impossible. Recovery pros say it bluntly: if you care about the data, don’t run chkdsk on a failing drive.

And even if it runs without issue—chkdsk still doesn’t fix a physical bad sector. It just marks it as bad and stops using it (remapping to spares is handled by the drive’s firmware, which is why the Reallocated Sector Count grows). The system may “stop crashing,” but the drive hasn’t healed. Rule from our shop: don’t run any “repairs” until your data is safe. We clone the drive first (a 1:1 bit-for-bit copy) and only work with the copy—we don’t touch the original unnecessarily.

Watch out with SSDs: repeated surface scans (chkdsk /r) on SSDs are pointless, and Microsoft itself warns that they needlessly eat up write cycles and shorten lifespan. Only scan an SSD in exceptional cases.

There are also specialized tools for HDD surface scanning with a visual map—like HD Tune, Victoria, or HDDScan. In the lab, we also use tools for controlled block-by-block drive imaging (PC-3000 and similar equipment). They show the drive as a grid: green blocks read fine, yellow ones are slow or require retries, red means a read error. Colors aren’t universal; always read the legend of the specific program.

Illustrative block-level read map based on lab diagnostics: green reads fine, yellow means slow/problematic reads, and red means an error. A few scattered blocks may not spell disaster; contiguous red areas are a serious signal.

Don’t confuse this with defragmentation. A defrag map shows where file fragments lie on the disk. A surface or block-level imaging map primarily shows latency and read errors—that is, whether specific spots on the drive can still be read reliably.

When to keep watching a drive—and when to toss it

In practice, it boils down to a simple decision tree:

• Zero to a few reallocated, stable for a long time → the drive can keep serving. Just monitor it (check SMART once in a while) and have a backup. It may easily run for years.
• Reallocated or pending sectors are growing → the drive is on its way out. Back up your data and plan a replacement so it doesn’t catch you off guard.
• Uncorrectable Sector Count > 0, or contiguous red areas on the map → you’ve likely already lost some data. Back up immediately and retire the drive. This isn’t the time to play repair.
• Drive is clicking, ticking, disappearing and reappearing in the system → power it off and don’t keep trying. Every spin-up worsens the odds. This is straight-up a job for data recovery.

Replacing a drive is cheap these days, and for a laptop or desktop it’s routine—see SSD/HDD replacement. It almost always pays off more than repeatedly reviving a drive whose firmware is running out of spare sectors. If you’re torn between repair, reinstall, and the end of the drive, we cover that too in when to repair, reinstall, or replace your computer.

SSDs are a slightly different story

An SSD has no spinning platter, so it doesn’t have “bad sectors” in the mechanical sense. But it has its own wear—NAND memory cells have a limited number of writes, and the firmware remaps bad cells too. So in CrystalDiskInfo for an SSD, watch values like Percentage Used / wear and data integrity errors instead.

And one thing you should never do to an SSD: don’t defragment it. Defragmentation just needlessly eats write cycles. SSDs are maintained by the TRIM command, and Windows handles it automatically—in the “Optimize Drives” tool, it runs defrag on HDDs and TRIM on SSDs, roughly once a week. The average user doesn’t need to do anything.

Fragmentation mainly slows down traditional HDDs. Red in a defrag map usually doesn’t mean a bad sector; it means fragmented files. SSDs don’t fragment in a meaningful way—and shouldn’t be defragmented.

What you can handle at home and when to call a shop

At home, you can comfortably:

• make a backup of important data—that’s always step one,
• check SMART (CrystalDiskInfo) and look at reallocated/pending sectors,
• on a healthy drive after an unclean shutdown, optionally run chkdsk C: /f for logical errors.

What you should not do at home: on a drive with growing bad sectors (or one that’s clicking/disappearing), don’t run chkdsk or “repair” tools until you’ve backed up your data. You could finish off a dying drive or overwrite data that could still have been saved—and then only data recovery can help.

It’s time for a shop when: reallocated sectors are growing, the drive is clicking or disappearing from the system, you can’t read from it—or you simply don’t want to risk your data and want it done right. At our PC service in Brno, we’ll measure the drive’s health, tell you straight whether it’s still worth using, and if things are bad, recover your data while it’s still possible.

Bottom line: only logical bad sectors can be “fixed”; physical ones can only be worked around. A drive where they’re growing isn’t a candidate for repair—it’s a candidate for replacement. And since nearly a quarter of drives die without warning, the most important rule stays the same—have a backup before you need one.