Is your drive failing? How to check SSD and hard drive health in Windows
Most drives warn before they fail. They count their own errors, wear and temperature, and report them to anything that asks — Windows asks, but rarely says anything. Here is how to read those numbers yourself, and what to do about them.
Updated 26 September 2026
Signs a drive may be failing
- Files that will not open or copy, or copies that stall partway.
- Long freezes when the PC opens or saves something, with the drive light on.
- Clicking or grinding from a hard drive, the kind with spinning disks.
- Disk errors in Windows’ own log. Open Event Viewer → Windows Logs → System and look for the source disk: event 7 says the drive has a bad block, and event 153 that a read or write had to be retried.
A slow PC on its own is rarely a failing drive. Why is my PC slow? covers the far more common causes.
Check it in Windows 11
For NVMe SSDs, Windows shows the drive’s own health report. Open Settings → System → Storage → Advanced storage settings → Disks & volumes, choose the drive and select Properties. The Drive health section gives the estimated remaining life, the available spare and the temperature, with a warning if the drive reports a problem. SATA SSDs and hard drives do not show this section; the next check covers them.
Check any drive with PowerShell
Right-click the Start button and choose Terminal (Admin). First, the status Windows keeps for each drive:
Get-PhysicalDisk | Format-Table FriendlyName, MediaType, HealthStatus, OperationalStatusHealthStatus should say Healthy. Warning or Unhealthy means Windows has flagged the drive. Then the drive’s own counters:
Get-PhysicalDisk | Get-StorageReliabilityCounter | Format-Table DeviceId, Wear, Temperature, ReadErrorsUncorrected, WriteErrorsUncorrected, PowerOnHoursWear is the share of its rated write endurance an SSD has used, as a percentage. The two uncorrected error counts should be zero. Not every drive reports every counter, and a blank is not a fault.
Other tools
CrystalDiskInfo, which is free and open source, reads each drive’s full SMART record and sums it up as Good, Caution or Bad. On a hard drive, Caution usually means reallocated or pending sectors: parts of the disk the drive no longer trusts. Drive makers’ own tools, such as Samsung Magician, WD Dashboard and Crucial Storage Executive, show the same and can also update the drive’s firmware.
What the readings mean
- Healthy, or Good, with no errors. Nothing the drive knows about is wrong. Keep backups anyway: drives can also fail without warning.
- Wear above about 80%. The SSD has used most of its rated writes. Many drives run well past 100%, but they grow slower at writing and eventually switch to read-only. Not an emergency; a reason to plan a replacement.
- Uncorrected errors, reallocated or pending sectors. The drive has been asked for data it could not return. Whatever was there is gone, and more usually follows. Back up now and replace the drive.
- A failure prediction, or Unhealthy. The drive expects to fail, and is usually right. Back up now and replace it.
- An NVMe SSD above about 70 °C. Fast SSDs slow themselves down when hot, which looks like a large copy that starts quickly and then collapses — easy to mistake for a failing drive when it is only a hot one.
What to do next
Replacing a failing drive
Copy your files off first, starting with whatever cannot be replaced. For the drive Windows runs from, installing Windows afresh on the new drive is the safer choice: cloning copies whatever the failing drive has already corrupted. Cloning a drive that is only worn, with no errors, is fine.
Cooling a hot SSD
- Fit the heatsink that came with the drive or the motherboard, if it is not already on.
- Keep the drive out from directly under the graphics card, which blows heat onto it, if the board has another slot.
- Improve the case’s airflow: M.2 slots often sit in still air.
How Bottleneck checks this
Bottleneck reads each drive’s health on every scan: the status Windows keeps, the drive’s own failure prediction, its uncorrected errors, its wear and its temperature. A failing drive goes to the top of the report, above every performance finding, because nothing else it could tell you matters as much. It warns when an SSD has used 80% of its write endurance, and when an NVMe drive runs hot enough to slow itself down.
Reading a drive’s own records needs administrator rights, which is one reason Bottleneck asks for them. Without those rights, it can still read the status Windows keeps.