Reallocated, Pending, and Uncorrectable Sectors: Read the Sequence, Not Just the Count
A health tool flags your hard drive Caution, and three numbers are highlighted: reallocated sectors, current pending sectors, and uncorrectable sectors. The usual next step — searching for a “safe” number — leads into a mess of conflicting answers, from “replace it at the first one” to “worry at five hundred.” Both extremes miss the point. These three counts are not three separate alarms. They are three stages of one process, and the stage your drive has reached, plus whether the numbers are moving, tells you far more than any single total.
The three counts are one story, told in order
A sector is a small unit of storage on the drive’s surface. When the drive struggles to read one, the story begins. First, the sector becomes pending: Current Pending Sector Count (attribute 197, shown as C5 in some tools) counts unstable sectors that have not yet been remapped. Pending means suspect, not yet convicted — the drive is waiting to see whether the sector can be read or written successfully.
If the drive later resolves a pending sector by moving the data to a reserved spare area, that sector becomes reallocated. Reallocated Sector Count (attribute 05) is the permanent record of confirmed bad sectors whose data was moved to the spare pool. A reallocation where the data survived is the drive’s repair mechanism working as designed.
The third stage is the one that matters most. If the data in a bad sector could not be recovered during that process, it also counts as uncorrectable — Uncorrectable Sector Count (attribute 198, C6) and its close relative Reported Uncorrectable Errors (attribute 187). An uncorrectable count above zero means data was already lost, whether or not any program has complained yet. That is the difference between the stages: pending means data is at risk, reallocated means data was rescued and moved, and uncorrectable means the rescue failed.
Why there is no safe number
Three facts make a universal threshold impossible. First, tools show two different values for each attribute: a manufacturer-scaled “normalized” value and the raw count. A reader who sees “100” beside Reallocated Sectors Count may be looking at the normalized health scale, not one hundred bad sectors — comparing that number against a forum threshold for raw counts produces pure confusion. Always compare raw value against raw value.
Second, raw values are vendor-specific in how they are packed and reported, and even the meaning of the pending count is described by repair references as highly dependent on the manufacturer. A threshold borrowed from a different brand’s drive is a borrowed answer.
Third, the operators who manage drives at the largest scale do not use a magic sector total. Backblaze, which records SMART data from its data-centre fleet, uses five attributes as evidence of impending failure — 5, 187, 188, 197, and 198 — chosen because they are consistent across manufacturers and good predictors of failure, out of more than seventy available statistics. The working question in that method is whether the raw values are above zero and how they change, not whether a count has crossed an arbitrary line. A small count that is growing describes a deteriorating surface. The same count, unchanged for a year, describes an old incident. The number alone cannot tell those two drives apart; the trend can.
The decision table: what each combination means
| What the raw values show | Reading | Action |
|---|---|---|
| 05 = 0, 197 = 0, 198 = 0 | No sector problems recorded | Normal backups; recheck on your regular schedule |
| 05 above 0, stable across two checks a week apart; 197 = 0, 198 = 0 | Past damage, repaired by reallocation, not currently progressing | Back up, keep the drive in service for non-critical use only if you accept the risk, and keep watching the trend |
| 197 above 0, any value of 05 | Sectors the drive cannot currently read reliably are waiting for a verdict | Back up now, starting with the files you cannot replace; plan replacement — a rising pending count is a replace signal in vendor guidance |
| 05 growing between checks | The surface is actively deteriorating | Back up now and replace; growth, not the total, is the failing signature |
| 198 or 187 above 0 | Data has already been unrecoverable at least once | Back up what remains immediately and replace the drive as soon as possible |
| 188 (Command Timeout) growing, sector counts at 0 | Commands are timing out without sector damage — check the connection path before condemning the drive | Back up, then test with a different cable and port before buying a replacement |
Note what the table never does: it never declares a raw reallocated total “safe.” New drives should show zero reallocated sectors, and an occasional one or two on an older drive usually does not represent a serious problem on its own — but “usually” is doing real work in that sentence, and only the trend check below turns it into a decision.
The two-check trend procedure
- Back up before you diagnose. If any of 05, 197, or 198 is above zero and the drive holds anything irreplaceable, copy that data to a different drive first. Diagnosis involves reading the whole surface, which is exactly the workload a weakening drive handles worst.
- Record the raw values, with the date. Open your health tool — CrystalDiskInfo on Windows, or smartctl -A with smartmontools — and write down the raw values for attributes 05, 187, 188, 197, and 198, plus power-on hours. A screenshot saved with the date in the filename does the same job.
- Use the drive normally for about a week. Do not run surface scans, repair passes, or defragmentation in the meantime; they add stress and can move the counts for reasons that muddy the comparison.
- Record the same five raw values again. Compare like with like: raw to raw, same tool, same drive.
- Apply the decision table to the pair of readings. All five unchanged points to monitoring. Any growth in 05, 197, or 198 points to replacement. A jump in 188 alone points to the cable and power check first.
What success looks like: you hold two dated readings, and you can state in one sentence which row of the decision table your drive sits on — stable, progressing, or already losing data — instead of arguing with a threshold from a forum post about a different brand.
The look-alikes: when the cable, not the surface, is talking
Command Timeout (attribute 188) earns its place in the five-attribute set, but it is also the attribute most likely to be blaming the wrong component. A timeout records that a command took too long or was aborted; a failing SATA cable, a loose power connector, or a struggling controller can produce exactly that without the disk surface being damaged at all. The separating test is the combination: timeouts climbing while all three sector counts stay at zero points outward, at the connection. Timeouts climbing together with pending or reallocated sectors points inward, at the drive. Two other attributes cause regular false alarms and should never start a panic on their own: Raw Read Error Rate (01) and Seek Error Rate (07) carry vendor-packed raw values on some brands that look enormous while describing normal operation, which is why general references warn that their raw values are manufacturer-dependent and difficult to interpret without the maker’s own diagnostic tool.
What not to do to a drive with pending sectors
The Caution status cannot be repaired away. It reflects physical wear recorded by the drive itself, and no software resets those counts to zero — a tool claiming to “fix” reallocated sectors is at best hiding the report. Running disk-repair utilities against a drive that is actively failing can make it worse, because repair passes force repeated reads of the weakest areas. The data-safe order is always the same: copy what matters to a healthy drive, in order of irreplaceability, then — and only then — run diagnostics, self-tests, or warranty checks. If the drive holds the only copy of irreplaceable data and it is clicking, freezing, or disappearing from the system, stop experimenting and consider a professional data-recovery service; this guide is general information, not a recovery procedure for a drive in mechanical distress.
Questions owners ask about sector counts
Is one reallocated sector a death sentence? No. One or two, stable over time, with pending and uncorrectable at zero, is an old repaired incident on most readings. It becomes serious when it grows, or when it arrives with company — pending sectors waiting or uncorrectable errors recorded.
CrystalDiskInfo says Caution but everything works. Is the tool exaggerating? Caution is commonly triggered by exactly these counts rising above zero. The drive working today and the surface deteriorating are both true at once; that gap is the whole value of the warning, and the reason the backup comes before the debate.
Pending sectors disappeared at the next check. Fixed? Disappeared pending sectors were resolved one way or the other: reallocated, which raises count 05, or read successfully after all. Check which happened. Pending falling while reallocated rises by a similar amount is the sequence completing, not a clean bill of health.
Should a drive with any uncorrectable sectors stay in a backup rotation? No. A drive that has already failed to recover data once has demonstrated the exact failure a backup drive exists to survive. Retire it from backup duty even if it still formats and appears to work.