Data error (cyclic redundancy check): what it means and what to do with the drive

The short answer: „Data error (cyclic redundancy check)" does not mean that a file is corrupted in the sense that its contents contain a typo—it means the file could not be read consistently. The checksum stored with the data did not match what the drive actually returned. On a hard disk, the most common causes are a bad sector, a failing platter surface, or a faulty data path; on SD cards and flash memory, the cause is more likely to be a worn-out cell or controller. What you do next matters: every additional copying attempt forces the drive to read the exact same problem area again and again—and on a mechanically damaged drive, that reduces the chances of recovery. Stop copying, disconnect the drive, and perform any further work on a sector-by-sector image, not the original.

This is one of the most common error messages people bring to us. It appears halfway through copying a folder of photos, when opening an accounting database, while extracting a backup, or when trying to move data from an external drive to a laptop. It looks like a minor issue—you can close the dialog and try copying again. But repeating the attempt is the most dangerous thing you can do at that point.

What a cyclic redundancy check actually is

CRC (cyclic redundancy check) is a checksum. When data is written, a short number is calculated from it and stored alongside it. During reading, the calculation is repeated and the two numbers are compared. If they do not match, the system knows that what it has just read is not the same as what was originally written—and refuses to return it, because silently substituting bytes would be worse than displaying an error message.

Windows has a dedicated system error for this condition: number 23 (0x17), ERROR_CRC, with the message Data error (cyclic redundancy check). You may also encounter it as 0x80070017, which is simply the same error 23 written in the format used by installers and updates.

Checksums exist at several layers at once: inside the drive for every sector, on the SATA or USB bus during transmission, and sometimes in the file format itself, typically in ZIP or RAR archives. The message alone therefore does not tell you exactly where the problem occurred—it only tells you that somewhere between the platter and your application, the data no longer matched its checksum.

What it means on a hard disk

With a conventional HDD, the cause is usually one of three things, and no Windows dialog can distinguish between them:

• Bad sectors and surface degradation. The magnetic layer on a platter does not last forever. A sector that could be read without any problems a year ago may return unreadable data today. The drive tries to reconstruct it using its own error correction, and if that fails, it returns a CRC error. The problem often develops gradually: first one file, then an entire folder a month later.
• Mechanical failure. A damaged read head or a partially contaminated surface looks exactly the same from the outside—only „some files cannot be read." The difference is that every additional read can change the condition of the medium.
• The data path. A faulty USB cable, a loose SATA connector, inadequate power for a 3.5″ external drive, a defective USB controller in the computer, or an overheated adapter. In this case, the drive is often not damaged at all, and the data can be copied without difficulty after replacing the cable.

The last case is the only one where it makes sense to „try" something—and even then, restraint is essential. If the drive has not been dropped or exposed to liquid and sounds no different than before, try once with another cable known to work and a different port. For 3.5″ models such as the Seagate Expansion Desktop or WD Elements Desktop, you can also try another power adapter with the correct voltage and polarity. Once. If the drive clicks, grinds, repeatedly spins up, or disappears from the system, further attempts will reveal nothing and only make its condition worse.

It means something different on SD cards, USB flash drives, and SSDs

Flash memory has no platters or heads, but it still uses checksums. When a cell loses its charge after many write cycles or after sitting unpowered for a long time, the controller may be unable to reconstruct the data and reports the same error.

Flash media have another important characteristic: repeatedly connecting and reading them can trigger the controller’s own clean-up processes. What „only" wears the surface of a hard disk may cause the state of flash memory to change by itself between attempts.

Why repeated copying attempts cause harm

When copying ends with a CRC error and you select „Try again," the computer returns to the same location and begins reading the same sector repeatedly. The drive is not idle during this process: it makes repeated attempts to read around the problem area, recalibrates itself, and moves its heads back and forth. Nothing dramatic may happen on a healthy drive with a single isolated bad sector. But on a drive whose surface is deteriorating or whose head is already failing, one unreadable area can turn into ten.

There is also a second effect that people do not expect. Starting the copy „from the beginning" means repeatedly reading all the data that is still intact. The drive therefore spends most of its remaining life reading files that could already have been safely recovered, reaching the problem areas in a more exhausted state every time.

That is why laboratories do not work with individual files but with a sector-by-sector image: reading is controlled, the number of retries in problem areas is limited, damaged regions are skipped, and they are revisited only at the end, once the rest of the data is safely stored elsewhere.

Why CHKDSK /r puts data on such a drive at risk

The most common advice online is to “run chkdsk /f /r.” When a drive is reporting CRC errors, this advice often causes people to bring it to us in a substantially worse condition than before.

According to Microsoft’s documentation, the /r parameter locates bad sectors, attempts to recover readable information from them, and—where possible—replaces bad sectors. /r also includes the functionality of /f, which repairs file-system errors. These are two write operations on a medium that has just demonstrated that it cannot read reliably:

• Writing to the file system. CHKDSK repairs the NTFS structure based on whatever it manages to read at that moment. If some metadata is read incorrectly, the „repair" may remove references to files that could still have been reconstructed from a sector-by-sector image. The result is often a FOUND.000 folder containing numbered fragments instead of the original names and structure.
• Checking every sector. /r reads the entire surface. On a mechanically failing drive, this means hours of the most demanding operation possible—exactly what you want to avoid.

We discuss when CHKDSK is and is not appropriate in a separate article, CHKDSK on an external drive: when it helps and when it causes harm. In short, it should not be the first step on a drive containing data that is not stored elsewhere. If used at all, it should be the last step—and only on a copy.

The correct procedure, step by step

1. Stop copying. Close the dialog, do not try „one more time," and do not run another copying tool on the same folder.
2. Check the data path—once. Try another high-quality cable and another port, ideally one directly on the motherboard rather than through a hub. For 3.5″ models, also check the power adapter. If this does not resolve the error, the data path was not the cause.
3. Disconnect the drive and leave it powered off. Do not run CHKDSK, format or initialise it, perform any „repair„ offered by the system, or connect the drive „just for a moment" to see whether it happens to work today.
4. Have a sector-by-sector image created. The aim is to move as much data as possible from the failing medium to a healthy one using controlled reading and a map of problem areas.
5. Recover from the image, not the original. File-system reconstruction, lost-partition searches, and file-integrity checks should all be performed on the image. The original should no longer be touched.
6. Check the recovered files. This is particularly important for databases, archives, and video: a file with the correct size is not necessarily usable.

If you are unsure which stage you have reached, leave the drive powered off and contact us or call our data recovery line at 775 556 063. We perform diagnostics in our own laboratory in Brno; we also explain separately when it makes sense to bring the drive in person and when shipping it is sufficient.

Three situations where the message can be misleading

Not every CRC error means a drive is dying—and conversely, not every apparently „resolved" case is truly resolved. These are the three situations we encounter most often:

• The error appears with only one particular file and nowhere else. This is often caused by an isolated bad sector in the area containing part of that file. The rest of the drive may be fine—but this is a warning, not a coincidence. Back up the remaining data elsewhere and stop trusting the drive as your only storage location.
• The error disappears after replacing the cable or switching to another port. That is excellent news, but follow the issue through to its conclusion: if the data transfer was failing, data previously copied „successfully" may also have been corrupted. For archives, backups, and databases, verify the checksums rather than merely checking that the file opens.
• The error occurs in a different place every time. This is the worst scenario. It usually means the surface is deteriorating or there is a problem with the heads, and time is working against you. Do not experiment—the drive should be taken out of service and sent for diagnostics.

How to avoid the same problem in the future

Most CRC-related recovery jobs begin the same way: the only copy was stored on a single medium that reached the end of its life. Prevention is not difficult; it is simply easy to overlook while everything is working.

• Follow the 3-2-1 rule: keep three copies of your data, on two different types of media, with one copy off-site or in the cloud.
• Test restoration, not just backup creation. A backup from which you have never restored anything is only a hope.
• Monitor SMART data. Increasing counts of reallocated and pending sectors can provide weeks or even months of warning.
• An external drive is not an archive. Portable drives and cards spend their lives in backpacks, cars, and near the edge of desks—take that into account when deciding where to keep „the only copy."

When the drive needs a laboratory immediately

• the drive clicks, grinds, squeals, or repeatedly spins up,
• CRC errors increase between attempts—one file today, an entire folder tomorrow,
• the drive disappears from the system during copying,
• it reports an implausible capacity or appears as RAW,
• it was previously dropped, exposed to liquid, or connected during a power surge,
• it contains the only copy of accounting records, client work, or a company database,
• CHKDSK has already been run and files have disappeared or a FOUND.000 folder has appeared.

What professional diagnostics can determine

The purpose of diagnostics is not simply to confirm that „the files cannot be read." It must distinguish between a faulty cable or bridge, defective electronics, damaged firmware, worn read heads, and a degraded surface—each cause requires a different approach.

In the laboratory, we check the exact drive model and revision, power consumption during start-up, firmware identification, the condition of individual heads, and sector readability. Tools such as PC-3000 and DeepSpar make it possible to control reading in ways an ordinary computer cannot: limiting the number of retries on a problem sector, reading with individual heads, skipping a damaged area, and returning to it only at the end. The goal is to create the most complete working image possible if the condition of the medium allows it—with extensively damaged drives, producing a usable image may not be possible.

The price is determined not by the brand or capacity but by the type of failure and the number of hours of controlled reading required. You can find indicative price ranges by fault type in our article on how much hard-drive data recovery costs; no one can provide an exact figure before diagnostics.

Frequently asked questions

What does a data error (cyclic redundancy check) mean on an HDD?

It means the data could not be read consistently—the checksum stored with the sector did not match what the drive returned. On a hard disk, this is commonly caused by bad sectors, a degrading surface, or mechanical failure, but a faulty USB cable, inadequate power, or a defective controller can produce exactly the same symptom. Only diagnostics can distinguish between them, not repeated copying attempts.

Could it work if I try copying it for the tenth time?

Sometimes it does, but that is the most expensive way to find out. Every attempt forces the drive to read the exact location causing the problem, and on a mechanically damaged drive, the damage can spread with every pass. If you genuinely need the data, it is safer to move everything readable into a sector-by-sector image first and deal with the problem areas only afterwards.

Can CHKDSK eliminate CRC errors?

CHKDSK repairs the file-system structure, and with /r it also checks the surface and attempts to replace bad sectors. On a healthy drive with an isolated logical error, it is a legitimate tool. On a drive that has just reported CRC errors, however, it writes to an unreliable medium—and if the metadata is read incorrectly, the „repair" may remove references to files that could otherwise still be recovered.

The error also appears on an SD card from my camera. Is it the same thing?

The symptom is the same, but the cause is different. On memory cards, the most common causes are worn-out cells, an interrupted write when the camera battery ran out, or a counterfeit card that falsely reports its capacity. Remove the card, do not write anything to it, and do not format it—our guide explains how to proceed with SD card data recovery.

Does a CRC error in a ZIP or RAR archive mean the drive is damaged?

Not necessarily. Archives have their own checksums, so they report the same error if the file was corrupted during downloading or network transfer. However, if the same archive repeatedly fails on the same drive and additional files become unreadable, the signal is coming from the medium, not the archive.

Can I create a sector-by-sector image myself?

If the drive is otherwise healthy and has only a few unreadable areas, this may be feasible. If it clicks, disappears from the system, or deteriorates between attempts, DIY cloning is the fastest way to turn a recoverable case into an unrecoverable one—ordinary tools cannot limit repeated read attempts or read through individual heads.

Sources

• Microsoft: System Error Codes (0–499)—ERROR_CRC 23 (0x17)
• Microsoft: chkdsk—parameters /f, /r, /b
• Microsoft: Windows File Recovery