smartctl Output Interpreter

How this reads your output

Paste the report and the attribute table is read into a common shape, then judged. The judgement is deliberately narrow. A drive reports thirty attributes and most of them predict nothing: Backblaze publishes failure statistics across hundreds of thousands of drives and finds five that genuinely correlate with failure. Those five carry the verdict here. Pending sectors are treated as the most urgent, because they describe data that cannot be read right now rather than damage the drive has already contained. Interface errors are separated out entirely, since they are almost always a cable and telling someone their drive is failing when the fix costs a few pounds is the most common wrong answer in this category.

What the results mean

Current pending sectors (197)

Sectors that failed to read and are waiting to be remapped. Unreadable now — this is data loss in progress, and the most urgent value on the list.

Reallocated sectors (5)

Sectors that failed and were remapped to spare area. Damage the drive has contained. A stable count is a drive that took a knock; a growing one is a drive on its way out.

Command timeout (188)

Commands that did not complete. Frequently the cable or the power connector rather than the drive — worth eliminating before condemning anything.

UDMA CRC errors (199)

Corrupted transfers between drive and controller. Almost always a bad data cable. The counter never resets, so what matters is whether it rises after the cable is changed.

Normalised value versus raw

The normalised value is a vendor scale where higher is better and the threshold is the failure floor. The raw value is the actual count, and for the five predictive attributes it is the number that matters.

Common problems and fixes

No attribute table was found

Paste the whole report. A drive in a USB enclosure or behind a RAID controller often reports no table at all, which is the enclosure rather than the drive.

Attributes show huge raw values but the drive seems fine

Check which attribute. Seek error rate and raw read error rate are vendor-encoded and read in the millions on healthy drives — they are noise, not evidence.

The overall self-assessment says PASSED but sectors are reallocated

Both are true. The self-assessment only fails once a vendor threshold is crossed, which is late. Reallocated sectors are worth watching well before that point.

The count went up since last month

That is the finding. A rising count on any of the five predictive attributes means active degradation — plan the replacement rather than waiting for the drive to fail.

Frequently asked questions

More in Interpreter

• SMART Data Interpreter (CrystalDiskInfo)Paste a CrystalDiskInfo report and get the same verdict, from the same five attributes.
• Windows Stop Code / BSOD LookupLook up a Windows blue-screen stop code and get the likely cause and next step.