How to Run Windows Memory Diagnostic in Windows 11

Windows 11 includes Windows Memory Diagnostic for checking system memory outside the normal desktop session. Microsoft documents it as a diagnostic step when memory problems may contribute to crashes or stop errors. It requires a restart because the test runs before Windows returns to the normal sign-in environment.

A clean result is useful evidence, but it does not prove that every DIMM, memory-controller setting, motherboard trace, memory profile, overclock, or workload is stable under every condition.

Save open files and close applications. Press Windows + R, type mdsched, and select OK. Microsoft documents this command as the entry point for Windows Memory Diagnostic.

Choose Restart now and check for problems when you are ready. The machine leaves the normal Windows session and starts the memory test automatically after restart.

Windows Memory Diagnostic runs its Standard test mix by default. For a normal first pass, let it complete without interruption. Duration varies with memory capacity, platform speed, and the selected test mix.

When the test finishes, Windows restarts automatically. Sign in normally. Microsoft says a notification with the result should appear in the notification area, although it can take a few minutes.

If Standard reports no problem but memory remains a credible suspect, Microsoft documents an Extended test mix. While the diagnostic is running, press F1, change Test Mix from Standard to Extended with the arrow keys, and press F10 to apply the setting.

Extended testing is more thorough and takes longer. It is an escalation within the built-in diagnostic, not proof that memory is the cause of a crash.

If the result notification does not appear or is dismissed, Microsoft documents the result in Event Viewer. Right-click Start, open Event Viewer, then navigate through Applications and Services Logs > Microsoft > Windows > MemoryDiagnostics-Results > Debug and inspect the recorded test entry.

Microsoft's current support guidance identifies Event ID 2001 as meaning that no memory errors were detected. Read the event message from your own run rather than relying only on an event number copied from another system.

If the diagnostic reports an error, stop treating the machine as known-stable. Return CPU and memory tuning to documented defaults if necessary, preserve the result, and follow the system or motherboard vendor's supported memory troubleshooting procedure.

A memory error does not by itself identify one DIMM as physically defective. The failure condition can involve a module, slot, memory-controller behavior, firmware, electrical margins, or an unstable configuration. Controlled testing at documented defaults is more informative than changing several variables at once.

A clean Standard or Extended run means this diagnostic did not detect a memory error during that test. It does not guarantee stability in every workload or configuration. Intermittent or configuration-dependent instability can escape a bounded test.

If crashes or corruption continue and still correlate strongly with memory, preserve the result and compare behavior at documented firmware defaults. Follow the motherboard or system manufacturer's supported isolation procedure for the exact platform.

A practical workflow is to save your work, run mdsched, allow Standard to complete, read the post-reboot result, use Extended when a deeper built-in pass is justified, and preserve the result alongside the original symptom and system configuration.

The tool provides a repeatable Windows memory check with a recorded result. Use that evidence to narrow the next step rather than turning either a pass or a failure into a broader conclusion than the test supports.