For an HDD, CrystalDiskInfo can display Caution when the raw value of a monitored sector-related attribute exceeds its configured warning threshold. The important next step is to protect the data and inspect which field triggered the warning. The label alone does not tell you when the drive will fail.
This guide focuses on HDD attributes 05, C5, and C6. SSD and NVMe health indicators need their own model-specific interpretation. If your device is an SSD, do not assume that every HDD attribute number has the same meaning.
Secure the files before trying to change the warning
If important files exist only on this drive and it is still accessible, copy the highest-priority files to a different physical device or backup service. Check that the copies open. Moving them to another partition on the same HDD does not protect against failure of that HDD.
If the drive repeatedly disconnects, reads fail, or it makes unusual mechanical sounds, stop repeated attempts and seek recovery advice for irreplaceable data. A long surface test should not be the first activity on an unstable drive holding the only copy.
Do not change CrystalDiskInfo’s health thresholds to make Caution disappear. That changes when a warning is shown; it does not improve the drive’s recording surface or recover unreadable data.
Identify the attribute behind Caution
Select the intended drive through CrystalDiskInfo’s Disk menu. Confirm the model, serial number, capacity, and drive letters. Use Function > Refresh and save both a screenshot and an Edit > Copy text export.
Read the attribute name together with its ID:
| HDD attribute | What it commonly describes |
|---|---|
| 05 — Reallocated Sectors Count | Sectors the drive has replaced using spare space. |
| C5 — Current Pending Sector Count | Unstable sectors awaiting further evaluation after a read problem. |
| C6 — Uncorrectable Sector Count | Sectors associated with errors the drive could not correct. |
Use the manufacturer’s definitions for the model when interpreting the exact counting behavior. The three values are not automatically disjoint sets that you can add to obtain a total number of damaged sectors.
Keep Current, Worst, Threshold, and Raw Values distinct. A normalized Current value of 100 is not a raw count of 100 bad sectors. CrystalDiskInfo’s configurable Caution threshold is also not the same thing as every manufacturer’s normalized failure threshold.
What each change can tell you
Reallocated sectors have already been handled by the drive
A nonzero 05 value indicates reallocation activity in the drive’s history. Reallocation lets the device use spare storage for affected locations; it does not establish that every earlier read succeeded or every file is intact.
Record whether the value is stable or rising. A new increase is information to bring to the manufacturer, especially when accompanied by read failures. A stable value does not certify future reliability.
Pending sectors are unresolved evidence
C5 can change as the drive evaluates the affected locations. A later reduction needs context: record whether the drive was read, written, tested, or repaired between snapshots.
Do not deliberately overwrite data just to make the pending count fall. If C5 becomes zero, check what happened to 05, C6, the files, and the symptoms before concluding that the problem is gone.
Uncorrectable errors deserve attention even if files still open
A C6 indication does not tell you which file, if any, occupies the affected location now. Some files opening successfully does not test the whole drive.
Preserve accessible data and the report. If you need further diagnosis, provide the exact model and the full attribute table so support can distinguish the drive’s reporting conventions from a generic interpretation.
Compare two records without losing the number format
Keep the software version, firmware, collection time, and raw-value format with each record. Raw values may be shown in hexadecimal; a number displayed after a format change must be converted consistently before comparison.
The table below is a made-up example using decimal counts. It is not a tested drive:
| Field | Check A → Check B |
|---|---|
| 05 | 4 → 5 |
| C5 | 2 → 1 |
| C6 | 0 → 0 |
| Recorded activity | Ordinary use between checks; no controlled test. |
Here, 05 rose by one and C5 fell by one. Those changes are consistent with a change in how affected sectors were handled, but the table alone cannot prove exactly which location changed or whether a particular file was recovered. The unchanged C6 value does not cancel the other observations.
Use the example as a recording method, not an acceptable-count threshold. There is no “safe number of pending sectors” established by this table.
Decide what to do with the record
| Finding | Next action |
|---|---|
| Caution with accessible files | Verify the backup, retain the report, and review model-specific support and replacement guidance. |
| Any relevant value rises between comparable records | Preserve both records and the activity between them. Ask the manufacturer about diagnostics or replacement. |
| C5 falls but another field rises | Keep the whole change; do not report only the improved value. |
| Read errors or repeated disconnects continue | Limit further testing and settle the data-recovery or backup plan first. |
| The only changing field is C7 | Investigate the SATA communication path separately instead of treating it as the same sector count. |
For a backed-up, stable drive, a supported vendor diagnostic can add evidence. Save the test name, options, date, and final result. Check whether a proposed option only reads data or attempts repair before accepting it.
If C7 is the field that keeps increasing, see C7 Interface CRC Error Count Keeps Rising: What to Check.
If Windows also logs bad-block events, keep their source, device path, and time. Compare them with this same HDD’s identity. A change in the health label does not explain the operating-system events by itself.
For that comparison, see Disk Event ID 7 but CrystalDiskInfo Says Good: What to Check.
The finished record should let you show which attribute triggered concern, how its value changed, and whether the data is protected. You do not need to force the label back to Good before deciding that a drive should be replaced or professionally assessed.
Leave a Reply