SSD Health Check: Reading Your Drive's Own Numbers

By LK Wood IV · 2026-09-18 · ~9 min read · St. Louis County, MO

A drive that is about to fail rarely announces it. The machine gets slower, a file refuses to copy, a game stutters where it never did, and you start wondering whether the SSD is on its way out. The good news is that every modern NVMe drive keeps a running log of its own condition, and Windows 11 will show you the important parts without installing anything.

I read that log on my own boot drive while writing this page. Below are the fields, what each one means, the thresholds that matter, and my drive’s actual numbers next to them so you have something real to compare against.

The 60-second check, no download

Open Settings, then System, then Storage. Scroll to Advanced storage settings, open Disks & volumes, select your SSD and choose Properties. Microsoft’s support page for storage critical warnings documents that exact path. The faster route is Win+R and ms-settings:disksandvolumes, which Microsoft lists in its ms-settings URI reference as the shortcut to that page.

Under Drive health you get three values, and Microsoft defines each one:

• Estimated remaining life — “The approximate percentage used, which is based on the manufacturer’s prediction of drive life.”
• Available spare — “A normalized percentage (0% to 100%) of the remaining spare capacity available.”
• Temperature — the whole device’s temperature in degrees Celsius, per Microsoft’s definition.

Microsoft’s own wording for the first field is confusing, so it is worth being plain about what you are seeing. The drive reports its wear as a percentage used. Settings shows you the other side of the same figure, as life remaining. A drive reporting 2% used is the drive Settings labels 98% remaining. Same estimate, subtracted from 100.

One caveat decides whether you see anything at all. Microsoft’s page states it in a single line: “Windows only monitors NVM SSD and not SATA SSD or hard disk drives (HDD).” If your drive is a 2.5-inch SATA SSD or a spinning disk, the section simply is not there.

That limit is worth knowing before you go looking for a fault. A Microsoft Q&A thread from March 2025 opens with a Windows 11 Home user reporting that “the whole section of device health is not showing.” The one substantive reply sent them to WMIC commands; the other asked where they were looking. Neither mentioned the documented reason the section can be absent.

What the fields mean

Windows shows you a summary. The drive itself keeps more, in a structure Microsoft documents as the SMART / Health Information log. These are the fields that answer “is it dying,” with Microsoft’s own definitions:

Two of those repay a second look.

Available spare is the one to watch, and the threshold is the reason. Spare capacity is the pool of reserve blocks the controller swaps in as cells wear out. The number on its own says little; its distance from the threshold says everything. Microsoft’s definition ties the two together: when Available spare falls below the threshold field, the drive can raise an asynchronous event. A drive at 100% spare has spent nothing. A drive at 12% against a 10% threshold is close to the line it will flag on its own.

Percentage used is an estimate, and the drive maker wrote the formula. Microsoft is explicit that the value is vendor specific, that it is allowed to exceed 100, and that it updates once per power-on hour. It is not a fraction of your TBW rating, which is a common misreading of the field. My own drive demonstrates the gap, below.

What my drive reports

This is the Samsung 970 EVO Plus 1TB that boots my desktop, read on 2026-09-18 at 14:09 Central. I have reviewed this drive for performance separately; this is its health log.

Now the part that changes how you read your own drive. Samsung’s data sheet rates the 1TB 970 EVO Plus at 600 TBW, and the warranty runs “for the stated time period or the TBW, whichever comes first.” My 51.03 TB of host writes is 8.5% of that rated budget. The drive’s own Percentage Used field says 2%.

Those two numbers disagree because they are answering different questions. The TBW figure is a warranty boundary Samsung published in 2021. Percentage Used is the controller’s live estimate of consumed life, built from what the NAND has actually endured. Neither is wrong. Treating one as the other is what produced the panic in the forum thread further down this page.

So read Percentage Used and Available Spare when you want to know how the drive is holding up. Save the TBW figure for warranty questions, which is where our SSD endurance calculator turns a daily host-write rate and a rating into years.

One number here is mine alone and should not be copied. Those 693 unsafe shutdowns across 1,214 power cycles work out at 57% of this machine’s power cycles. Microsoft defines the counter as incrementing when a shutdown notification is not received before power is lost, and that is the limit of what I can say about it, because what on this desktop triggers it remains an open question for me. It has produced no media errors in 5,615 hours.

What to do at each result

Everything clean. No critical warning, spare at or near 100%, zero media errors, Percentage Used low. Storage drops to the bottom of your suspect list. Note the numbers somewhere, because the trend over months tells you more than any single reading. If you arrived here from a blue screen, the CRITICAL_PROCESS_DIED triage picks up where a clean drive report leaves off.

Percentage Used climbing, spare still full. This is normal ageing. A drive at 40% used with 100% spare and no errors is reporting that it has consumed under half the life its maker predicted for it, and the SSD endurance calculator will tell you how long your write rate takes to spend the rest. Re-check quarterly and watch the rate of change rather than the value.

Available spare approaching its threshold, or any media errors. Copy your data off today. A spare pool that is draining means the controller is retiring blocks faster than the drive was designed for, and media errors mean data has already been lost that the controller could not recover. Neither improves.

Any critical warning bit set. Microsoft’s guidance is short: back up your data and contact support. Treat the drive as untrustworthy from that moment. If it is still under warranty, the health log is the evidence a support agent will ask for.

For the last two cases you need a replacement, and budget for more than you paid last time. NVMe prices rose sharply through 2026, and our reporting on the memory and storage squeeze has the numbers. Buy at least the capacity you are replacing, and check current listings before you commit to a model.

Some retailer links below are affiliate links; if you buy through them TechFuel HQ may earn a small commission at no extra cost to you. As an Amazon Associate I earn from qualifying purchases. Commissions never influence what gets recommended — see our disclosure.

If you want the comparison rather than a listings page, our NVMe SSD guide sorts drives by the role they fill.

Two routes that will waste your time

Older guides hand you a Command Prompt method: run wmic, then diskdrive get status, and read OK or Pred Fail. It no longer exists. Microsoft’s removal notice puts it in one sentence — “The WMIC utility has been removed from currently supported versions of Windows 11 and is no longer available as a Feature on Demand (FoD)” — with the change shipping in the August 2026 preview update for versions 24H2 and 25H2. On my 25H2 desktop the command returns 'wmic' is not recognized as an internal or external command.

The advice outlived the tool. That March 2025 Microsoft Q&A thread told its asker to run dism /online /Add-Capability /CapabilityName:WMIC~~~~ to install it first. As of Microsoft’s own notice, that capability is no longer there to install.

The same Status field still exists, so PowerShell gets it back in one line:

Get-CimInstance Win32_DiskDrive | Format-Table Model, Status, InterfaceType

Mine prints OK. That word carries almost no information — it is a pass/fail flag with no numbers behind it, which is why the health log above is worth the extra minute.

The second dead end is subtler. PowerShell has a cmdlet that looks perfect for this job:

Get-PhysicalDisk | Get-StorageReliabilityCounter | Format-List

Microsoft’s documentation for it describes counters covering “device temperature, errors, wear, and length of time the device has been in use,” and the page says nothing about needing administrator rights. Run it in an ordinary PowerShell window and you get:

Get-StorageReliabilityCounter : Access to a CIM resource was not available to the client.
    + CategoryInfo          : PermissionDenied

That is a permissions refusal rather than a missing capability, so the line wants an elevated Terminal. Microsoft’s page for the cmdlet never mentions it. The documentation gap is the trap, not the cmdlet.

Temperature: which number to act on

Your drive publishes two thresholds of its own. Mine declares a warning composite temperature of 82 C and a critical composite temperature of 85 C. It also keeps two clocks against them, which Microsoft defines precisely: one counts minutes at or above the warning figure while still below the critical one, the other counts minutes above the critical figure. Both read zero on my drive.

Samsung’s data sheet for the same drive gives a different figure: an operating range of “0°C to 70°C (Measured by S.M.A.R.T. Temperature Proper airflow recommended)”. So the firmware tolerates 82 C before it complains while the manufacturer specifies 70 C as the ceiling.

Design for the smaller number. The firmware thresholds exist to protect the drive from damage; the data-sheet range is where the manufacturer expects it to operate. A drive idling at 45 C like mine has plenty of margin. One sitting at 75 C during a long file copy is inside its firmware limits and outside its data-sheet spec, which is the point at which a motherboard M.2 heatsink earns its place.

Sustained heat also shows up in the log rather than only in the moment, so if the warning-temperature minute counter has moved off zero, the drive has spent real stretches of time above its own threshold whatever a single spot reading happened to say when you looked at it.

If your drive is SATA

Windows will not help you, by design. You need a tool that reads S.M.A.R.T. attributes directly, and CrystalDiskInfo is the free one I would reach for. Read its raw values rather than its headline percentage.

Crystal Dew World’s own criteria page shows why. It documents Caution when the raw values for Reallocated Sectors Count, Current Pending Sector Count or Uncorrectable Sector Count exceed their thresholds, and, for SSDs, Caution when “Remaining Life <= 10%”. Those are SATA-style attributes. The page does not document how the tool judges an NVMe drive, so on NVMe the fields in the table above remain the ones to read.

The gap between a tool’s summary and a drive’s data is not hypothetical. A September 2025 thread on the Tom’s Hardware Forum has a member reporting 10% health on a five-year-old M.2 SSD. One reply warned that CrystalDiskInfo “is generic software and doesn’t always interpret vendor specific SMART parameters properly,” and the owner came back to say the drive maker’s own utility rated the same drive in good health. A third member traced the 10% figure to a reported Percentage Used of 0x5A, which is 90. The tool read the field correctly. What it did with it afterwards was a judgement call.

The short version

Check the drive first. On NVMe that takes a minute in Settings and costs nothing, and it answers the question everything else only circles: read Available spare against its threshold, watch for any critical warning, confirm media errors sit at zero, and treat Percentage Used as the drive’s own estimate rather than a fraction of your warranty budget. Then write the numbers down. In six months the direction they moved will tell you far more than today’s reading does on its own.