Explainer

WHEA errors on Windows 11

What a WHEA error is, what the event IDs in Event Viewer mean, and what to do about one. Not every entry means your PC is broken.

Last updated: 14 November 2026 · applies to Windows 10 and Windows 11

The short version

WHEA is the part of Windows that records hardware faults reported by the processor, its memory controller and the PCIe bus. Some of those faults are corrected by the hardware itself and the machine keeps running unaffected; others are fatal and the machine stops. The entries show up in Event Viewer under a source called WHEA-Logger, each with a numbered event ID. Which number appears, and whether a crash came with it, is how you tell "an isolated blip" from "something is wrong".

What WHEA is

WHEA stands for Windows Hardware Error Architecture. It's how the processor and the motherboard firmware tell Windows that something physical went wrong: a bit flipped in a cache line, a PCIe link reported a transient error, the memory controller caught an error it could or couldn't correct. Windows writes each report into the System log with a stable source name and event ID, and that's what you can read after the fact.

You can look at them with Event Viewer directly: open eventvwr.msc, expand Windows Logs, click System, and filter by Source WHEA-Logger. If there are entries, they will tell you which kind.

The event IDs

  • Event 17: a corrected error on a PCI Express link. The event names a PCI Express root port, but the device behind it is the usual suspect: the graphics card, an NVMe drive, the Wi-Fi or network card. In Device Manager, View, Devices by connection shows which devices sit under that port. PCIe power saving (ASPM) on the device is a known trigger for floods of them.
  • Event 18: a fatal hardware error. Something went wrong that the hardware could not correct. This is normally paired with a crash or an immediate restart.
  • Event 19: a corrected machine check, usually reported against a processor core. The hardware spotted a problem and fixed it before Windows saw a consequence. A steady stream is a reason to look closer; on Ryzen, a Curve Optimizer offset or a memory fabric clock (FCLK) the chip can't hold is a known cause.
  • Event 47: a corrected memory error. Windows names memory as the component. On a machine with ECC memory, it is ECC fixing a flipped bit.

Other IDs exist, and what they mean depends on the component the record names. Each entry carries more detail on its General and Details tabs: the component, the error source, which processor the error came from, and the error code the hardware supplied. On a one-off event it is usually not worth decoding by hand; on a pattern it is where to start.

The BSOD: WHEA_UNCORRECTABLE_ERROR (0x124)

When Windows crashes with the blue-screen text WHEA_UNCORRECTABLE_ERROR and the hex code 0x00000124, that is WHEA reporting that a fatal hardware fault happened and nothing could be done about it. It is the most serious entry in this family. Even one is worth following up.

What usually causes them

On a desktop PC, roughly in the order they are easiest to rule out:

  • An unstable overclock or undervolt. A boost curve, a Precision Boost Overdrive or Curve Optimizer setting, an XMP or EXPO memory profile running faster than the memory controller can sustain, or a manual voltage offset that is just too aggressive for this silicon. Reverting to stock is the fastest test.
  • Heat. A hot CPU slows itself down rather than logging an error, but heat pushes a setting that was only just stable over the edge. Dust in the cooler, dried thermal paste or a case with no airflow can turn a working overclock into an unstable one.
  • Memory. Bad modules, a weak slot, or a stick not seated all the way down. An overnight run of MemTest86 is the standard test; Windows Memory Diagnostic (mdsched.exe) is a quicker first check.
  • PCIe link quality. A graphics card not fully seated, a poor riser cable or a worn slot can throw the link errors logged as event 17.
  • The power supply. A PSU that is too small, ageing, or failing to hold its 12 V rail when the graphics card's draw spikes can cause errors under load.
  • A degraded processor. A CPU that used to be stable at a given voltage and clock can stop being so, especially after heavy overclocking. Intel's 13th and 14th generation desktop chips are the documented case: in 2024 Intel traced their instability to elevated voltage and shipped microcode fixes in BIOS updates. The fix prevents the damage; it does not repair a chip that is already affected, and Intel extended the warranty on affected models.

When it isn't a problem

A single corrected entry after months of trouble-free running is rarely worth chasing. If entries started after a BIOS update, a driver update or a change to memory settings, that change is the first suspect.

A quiet event log with the odd corrected entry is normal. A log that fills with corrected entries over a few days, or any event 18 at all, is worth looking into.

Practical checklist

If you're seeing WHEA entries and want to work out whether to worry, roughly in this order:

  1. Open Event Viewer and note the event IDs and dates, and whether they started after a change: a BIOS update, a new memory profile, new hardware. For event 17, find the device behind the root port first.
  2. If anything is overclocked or undervolted, set it back to stock. XMP and EXPO count. Run for a week and see if the entries stop.
  3. Install the latest BIOS and chipset driver from your board or PC maker. On a 13th or 14th generation Intel desktop, make sure it includes Intel's 2024 microcode fix.
  4. Check temperatures under load. If the CPU is hitting its thermal limit, address that before going further.
  5. Run MemTest86 overnight. If it reports errors, test one stick at a time to find the bad one.
  6. If the entries continue on stock settings with healthy memory and temperatures, the PSU or the board is the next thing to suspect. Try a known-good PSU if you can borrow one.

In Clarity and Control

Clarity and Control both have a Hardware errors card under Tools. It reads the same WHEA-Logger entries Event Viewer shows and gives the corrected and fatal counts, with the event ID and time of the most recent ones. The card keeps no copy; the record stays in Windows' own log. If the Windows Event Log service is stopped, the card says so, with the reason, rather than reporting a clean log it never read.

Each result has a help page with the next step: