How to Diagnose Bad RAM: Windows Memory Diagnostic and Beyond

Faulty RAM is one of the most deceptive hardware problems a PC can have. It causes symptoms that look like software bugs — random crashes, BSODs, application freezes, corrupted files — but no amount of reinstalling Windows will fix a stick of RAM that’s gone bad. This guide walks you through every practical method to confirm whether your memory is the culprit.

Symptoms That Point to Bad RAM

Before pulling out tools, recognize the pattern. Bad RAM tends to produce:

• Blue screens of death (BSODs) with stop codes like MEMORY_MANAGEMENT, PAGE_FAULT_IN_NONPAGED_AREA, or SYSTEM_SERVICE_EXCEPTION
• Crashes that are workload-dependent — the system is stable at idle but crashes under gaming, video editing, or large file transfers
• Corrupted file errors when copying or writing large amounts of data
• Applications crashing randomly, especially memory-intensive ones like Chrome or Photoshop
• System refusing to POST or emitting beep codes on startup

If your crashes happen with a consistent stop code or only during heavy memory use, RAM is a prime suspect.

Step 1: Run Windows Memory Diagnostic

Windows ships with a built-in memory tester that requires no extra software. It won’t catch every subtle fault, but it finds obvious failures fast.

How to Launch It

Press Win + R, type mdsched.exe, and hit Enter. You’ll get two options:

• Restart now and check for problems — best if you can close what you’re working on immediately
• Check for problems the next time I start my computer — schedules a test for the next reboot

Choose the first option for immediate testing.

What Happens During the Test

Windows reboots into a blue diagnostic screen and begins writing patterns to every memory address, then reading them back. The default “Standard” pass runs two test passes. You can press F1 during the test to switch to Extended mode, which adds more aggressive tests at the cost of longer runtime (30–60 minutes depending on how much RAM you have).

Watch the Status section at the bottom. If hardware problems are detected, you’ll see a red message. After the test completes, Windows reboots to the desktop and logs results to Event Viewer.

Reading the Results

Open Event Viewer: press Win + R, type eventvwr.msc. Navigate to Windows Logs > System and look for a source of MemoryDiagnostics-Results. A result of “no errors detected” is good. Any mention of “hardware problems were detected” means you have a bad stick.

Step 2: Run MemTest86 for a Thorough Test

Windows Memory Diagnostic misses some error patterns that a dedicated tool like MemTest86 catches. If you got a clean result but still suspect RAM, run MemTest86.

Creating a Bootable USB

Download the free version of MemTest86 from the official site and use the included image writer (or Rufus) to flash it to a USB drive. Boot from the USB — you may need to change your boot order in BIOS/UEFI or use the one-time boot menu (usually F12 at POST).

Running the Test

MemTest86 starts automatically. The default configuration runs Test #1 through Test #13 in sequence. One complete pass can take 30 minutes to 2 hours depending on RAM size and speed. Running at least two full passes is recommended; intermittent faults sometimes only appear on the second or third pass.

Any red entry in the error table means defective RAM. Even a single error in an hour of testing is grounds for replacement.

Testing One Stick at a Time

If MemTest86 finds errors and you have more than one RAM stick installed, power down, remove all but one stick, and retest. Rotate through each stick individually to identify the specific module causing problems. This takes time but tells you whether you need to replace one stick or the entire kit.

Step 3: Physical Inspection and Reseating

Before assuming you need to buy new RAM, try reseating the modules. Oxidation on the contacts or a slightly loose stick can cause errors that mimic hardware failure.

1. Power off completely and unplug from the wall.
2. Ground yourself by touching the metal chassis or wearing an anti-static wrist strap.
3. Press down firmly on the retention clips at each end of the RAM slot — the stick should pop up slightly.
4. Remove each stick and inspect the gold contacts for green oxidation or physical damage. Wipe the contacts gently with a dry lint-free cloth or a pencil eraser (then wipe off any eraser residue).
5. Reseat the stick firmly until both clips click into place.
6. Reboot and rerun Windows Memory Diagnostic.

Step 4: Test in a Different Slot

Motherboards have multiple RAM slots, and the slots themselves can fail independently of the RAM. If reseating doesn’t help, move the stick to a different slot and test again. Consult your motherboard manual for the recommended single-channel or dual-channel slot configuration — populating the wrong slots can sometimes cause instability.

Step 5: Check XMP/DOCP Settings

RAM rated for 3200 MHz or faster ships at a default JEDEC speed (often 2133 or 2400 MHz) and relies on an XMP (Intel) or DOCP/EXPO (AMD) profile in BIOS to run at its advertised speed. Enabling XMP pushes the memory controller to its limits, and a slightly incompatible kit can become unstable.

Enter BIOS/UEFI (usually Delete or F2 at POST) and check whether XMP or DOCP is enabled. If it is and you’re having errors:

• Try disabling it and run at default JEDEC speeds. If crashes stop, your kit may not be fully compatible with your motherboard at XMP speeds.
• Try increasing DRAM voltage by 0.05V (e.g., from 1.35V to 1.40V) — some kits need a slight boost to run stably at advertised speed.
• Try lowering the XMP frequency one step (e.g., from 3200 MHz to 2933 MHz).

Interpreting Your Results

Test Outcome	Likely Conclusion	Next Step
Errors in one stick only	That stick is bad	Replace that stick
Errors in both sticks	Sticks bad, or motherboard slot issue	Test in different slots; replace if errors persist
No errors, but crashes continue	RAM is probably fine	Investigate CPU, storage, or PSU
Errors go away after disabling XMP	Compatibility or stability issue	Run at JEDEC speed or adjust voltage

When to Replace vs. RMA

If a stick tests bad and is still under warranty, contact the manufacturer for a replacement under RMA — most RAM brands offer lifetime warranties. Kingston, Crucial, Corsair, and G.Skill all have straightforward RMA processes online.

If the RAM is out of warranty and is a single-channel module in an older system, replacement DDR4 is inexpensive. A matched kit from the same manufacturer and specs is always preferable to mixing modules.

Final Notes

Bad RAM is annoying to diagnose because it mimics so many other problems. The workflow above — Windows Memory Diagnostic first, then MemTest86 for confirmation, then physical inspection and slot testing — covers the vast majority of cases. If all memory tests come back clean and you’re still chasing random crashes, turn your attention to the power supply, storage drive, or operating system.