S.M.A.R.T. attribute reference for HDD and SSD
A drive rarely dies without warning. It logs the warning first, in a set of numbers most people never open. We translate those numbers into plain English.
Seagate, WD, Toshiba, Samsung, Kingston, Crucial — one reference rail.
Self-Monitoring, Analysis and Reporting Technology is a set of counters and sensors built into almost every modern HDD and SSD. Each attribute tracks one thing: how many sectors were remapped, how hot the drive ran, how many hours it has spun, how much write endurance an SSD has consumed.
The drive stores a raw value, a normalized value, and a threshold for each one. When the normalized value drops to the threshold, the drive reports a failing status. That last part matters, because a drive can show alarming raw numbers for years and still pass, while another drive crosses a threshold quietly and fails within days.
Raw value
The drive's own count — the number people actually worry about.
Normalized value
A scaled score the firmware maintains, usually starting near 100.
Threshold
The trip wire. Cross it and the drive reports failing status itself.
A number only means something next to a trend.
A reallocated sectors count of 1 is common on used drives and usually stable. A count that was 1 last month and is 40 today is a different story. Check readings twice, ideally two to four weeks apart, before you decide to replace hardware. The calculator asks for a second set of values for exactly this reason.
Everything else in a SMART report matters less than these. Each one is explained in full, with plain-language thresholds, inside the attribute database.
Counts areas the drive already gave up on and replaced from spare space. A stable value is not a verdict; growth is. This is the attribute most often quoted in drive failure warning signs, and the most often misread.
Counts areas the drive cannot read right now and is still deciding about. If a later write succeeds the sector can clear; if it fails again it becomes reallocated. A growing pending count is one of the strongest early failure signals.
Exactly what it sounds like: total running time. Many desktop drives run well past 20,000 hours without trouble, so treat this as context for the other counters rather than a countdown clock.
The one reading that reflects your case, not just the drive. Sustained readings above 50°C deserve attention, and 60°C is a problem for both HDDs and SSDs. The vendor decides whether you see 194 or 190.
These are working reference points from vendor documentation and standard SMART practice — not a formula that predicts a date. Use them to decide when a reading deserves a closer look.
| Attribute | What it tracks | Comfortable | Watch closely |
|---|---|---|---|
| Reallocated Sectors (05) | Remapped bad areas drawn from spare space | Stable, single-digit on a used drive | Any jump between two readings weeks apart |
| Current Pending (197) | Unreadable areas still awaiting a decision | Zero, holding at zero on the next check | Anything non-zero, especially rising |
| Uncorrectable (187 / 198) | Reads and writes the drive could not verify | Zero under normal daily use | Recurring errors on the same files |
| Power-On Hours (09) | Total running time since the drive was made | Any age, if error counters stay flat | Old age plus newly moving error counters |
| Temperature (194 / 190) | Current drive temperature in Celsius | Broadly 25–45°C at idle | Sustained above 50°C; 60°C is a problem |
| Wear Leveling / Endurance | NAND write cycles consumed on an SSD | Normalized value still far from threshold | Steady normalized decline toward threshold |
Ranges are guidance, not cutoffs. Vendor firmware decides the normalized scale and the threshold each drive actually reports, which is why the same physical problem can look different between brands.
The same physical problem can appear under different names. Seagate and WD both expose reallocated sectors, but Toshiba reports some fields with different IDs on the MG Series, and SSD vendors replace surface-error logic with endurance logic entirely.
Crucial publishes its own SSD SMART guidance; Samsung surfaces much of the same information through Magician. Before you conclude anything from a raw value, confirm which vendor's mapping you are reading. The manufacturer notes page walks through those differences line by line — and flags the fields that are genuinely vendor-specific instead of guessing at them.
Read the manufacturer notes
Vendor coverage on this site
HDD lines emphasize surface errors. SSD lines emphasize write endurance. Both are covered, with the vendor differences called out.
You type in your raw values, it returns a plain assessment. No account, no upload, nothing stored. It weighs pending sectors and reallocated growth more heavily than power-on hours because age alone rarely kills a drive.
Reallocated and pending
Weighted heaviest, because they mark areas the drive already lost.
Temperature
Weighed for sustained heat, since it affects both NAND and platters.
Power-on hours
Read as context only. A long-running drive with flat counters is normal.
Three walkthroughs cover most real cases. Each one stays tool-agnostic — smartctl, CrystalDiskInfo, and vendor utilities all read the same underlying attributes, so the steps hold whatever you have installed.
For the machine that suddenly feels slow, clicks, or freezes mid-copy. Work through what the counters do first, what the drive's own self-test reports, and which symptoms point to the cable or the controller instead.
For a marketplace listing with no warranty. What to ask the seller to run, which endurance numbers to read, and why a low hour count does not replace a look at write volume.
Open the guideThermal limits explained for both NAND and platters, plus what sustained heat does to a drive that still reports healthy counters. Airflow, case placement, and which readings matter.
Open the guideA field like attribute 241 means "Total LBAs Written" on one Samsung drive and something unrelated on a Seagate. Generic attribute lists often skip this. Where the meaning is genuinely vendor-specific, it is labeled that way instead of guessed at.
Attribute definitions are compiled from vendor documentation and cross-checked against published SMART specifications. When a vendor changes a definition between firmware generations, both versions are listed. Sources and review practice are described on the About page.
About SMART Drive Health
If pending sectors are non-zero and growing, reallocated sectors jumped in one interval, or the drive fails its own self-test, back up first and ask questions later. Analysis is cheap; a failed read on a family photo library is not. Start with the backup, then use the calculator to decide the replacement timeline.
One honest limit: a clean SMART report means no tracked counter has crossed its warning zone yet. It does not mean the controller, the PCB, or the firmware is fine. SSDs occasionally fail with healthy-looking wear indicators. Keep a second copy of anything you cannot recreate.
Short answers to the readings that send most people looking. Longer explanations live in the database and the guides.
Attribute lookups, the calculator, and the guides link to each other on purpose. If you arrived with a specific reading already in hand, the database is the fastest route. If you only know the drive feels wrong, start with the dying-drive guide.
It means one sector was remapped, usually during manufacturing or an early write error. A single stable value is common and not automatically a reason to replace the drive. Watch whether it grows.
Treat it as a live warning. The drive cannot reliably read that area yet. If a later write succeeds, the sector can clear; if it fails again, it becomes a reallocated sector. A growing pending count is one of the strongest early failure signals.
There is no single cutoff. Many desktop drives run well past 20,000 hours. Age matters less than whether reallocated and pending counts are stable.
Sustained readings above 50°C deserve attention, and 60°C is a problem for both HDDs and SSDs. Brief spikes under heavy load are less concerning than a drive that idles hot.
Yes. SMART tracks wear and error counters, not every failure mode. Controller and firmware faults can occur while endurance indicators still look fine.
Vendors use their own attribute numbering and label the same behaviors differently. SSD lines emphasize write endurance; HDD lines emphasize surface errors. Always confirm the mapping for your specific vendor.