Signs Your RAM Is Failing and How to Test It Before It's Too Late

Your PC crashes three times in one afternoon. No warning, no error message you can act on—just a frozen screen followed by a reboot. You assume it's the graphics driver, or maybe Windows got corrupted again. You reinstall everything. It crashes again the next day.

Here's what almost nobody tells you: among all the components in a computer, faulty RAM is the hardest to diagnose because it rarely fails outright. It corrupts. It flips a single bit here, a byte there, and your system keeps running—badly. The signs of failing RAM look exactly like a dozen other problems, which is why so many people replace perfectly good drives and power supplies before finally testing their memory.

I've been through this more than once, on desktops and laptops, and I've learned the hard way that memory errors don't announce themselves. They hide. This guide shows you how to catch them.

Signs your RAM is failing (and why they're so easy to misread)

The frustrating truth: memory doesn't usually die. It degrades. A cell that once held a clean 0 starts returning a 1, and depending on what that bit was doing, you either notice nothing at all or your entire system face-plants.

In my experience, the tells cluster into a few recognizable patterns.

What are the symptoms of faulty RAM?

The most common symptoms of faulty RAM are random blue screens (especially with error codes like MEMORY_MANAGEMENT or PAGE_FAULT_IN_NON_PAGED_AREA), sudden reboots with no warning, applications crashing on launch, and files that save corrupted. But the ones people miss are subtler.

• Windows takes noticeably longer to boot than it did six months ago—and you've already ruled out the drive.
• A program that ran fine yesterday now throws errors on the exact same file.
• Your browser tabs crash randomly. Not all of them. Just some.
• Large file extractions fail partway through, repeatedly, at different points.
• The system won't POST at all—fans spin, screen stays black, sometimes with a beep code your motherboard manual can decode.

Notice how generic those sound? That's the problem. A dying hard drive produces almost identical behavior. So does an unstable power supply. So does thermal throttling on a dusty CPU cooler.

Which brings up the real skill here: not spotting symptoms, but proving it's actually the RAM.

The differential diagnosis, instead of guessing

When my desktop started rebooting under load, I burned a weekend chasing the wrong culprit. I reinstalled the graphics driver twice and swapped the PSU before a friend—patiently, over the phone—asked whether I'd tested the memory. I hadn't.

What I should have done first:

1. Rule out software. Run the built-in Windows Memory Diagnostic (search "memory" in the Start menu). It's fast and free. Non-conclusive, but it catches obvious faults.
2. Check the logs. Open Event Viewer and look for WHEA errors under Windows Logs. These hardware-level reports point toward memory more often than driver crashes do.
3. Test each stick alone. This is the step everybody skips, and it's the one that actually finds the bad module.

If you have two RAM sticks, pull one, run the machine for a day, then swap. If the crashes follow one specific stick into every slot, you've found your answer. If they persist regardless of which stick is installed, the fault may be in the motherboard's memory controller or the slots themselves—a different, uglier problem.

How to test RAM properly (MemTest86 and the quick options)

The Windows tool is convenient because it requires nothing but a reboot. MemTest86 is inconvenient because you have to build a bootable USB. The second one is worth the hassle, and I'll explain exactly why.

Is there a quick RAM test I can run right now?

Yes—Windows Memory Diagnostic. Type mdsched.exe into the Run dialog, choose to restart now, and let it work. It runs a basic pass and takes 15 to 30 minutes on a typical machine. It's a reasonable first filter.

What it won't do is push your memory hard. It runs a limited set of patterns, and a marginal stick can sail through it and still corrupt data under real load. I've had a module pass Windows Diagnostic twice and fail MemTest86 within the first twenty minutes. That experience is the whole argument for the heavier tool.

MemTest86: the test that actually finds marginal faults

MemTest86 runs outside your operating system, on bare metal, and hammers your RAM with patterns designed to expose weak cells. Here's the practical procedure:

• Download the image and write it to a USB stick with a tool like Rufus or Balena Etcher.
• Boot from the USB. You may need to change the boot order in BIOS/UEFI.
• Let it run. A single full pass takes one to several hours, depending on how much memory you have.
• If it reports even one error—one—stop. A healthy module produces zero errors across multiple passes.

Run at least two full passes if you get a clean first run. Overnight is ideal. Marginal faults sometimes only appear after the module has warmed up.

Test method	Time required	Thoroughness	Best for
Windows Memory Diagnostic	15–30 min	Basic	Quick first check
MemTest86 (1 pass)	1–3 hours	High	Confirming a suspected fault
MemTest86 (overnight, multi-pass)	6–10 hours	Very high	Marginal, intermittent faults
Module swap test	1–2 days	Real-world	Reproducing load-dependent crashes

Laptop-specifics: what changes when you can't just pull a stick

Laptops use smaller SO-DIMM modules, and on many recent thin-and-light models, the RAM is soldered directly to the board. Before you plan any swap, check your model's service manual. If memory is soldered and MemTest86 reports errors, you're looking at a motherboard-level repair rather than a $40 part.

The testing procedure is otherwise the same, with one caveat: laptops throttle hard, and a marginal module may only fail when the chassis is warm. Run the test with the machine plugged in and well ventilated, not balanced on a bedsheet.

Can RAM failure be fixed?

No—not if the module itself is physically failing. Dead cells on a memory chip can't be reflowed back to life. The fix is replacement. Full stop.

But here's the distinction almost every guide blurs, and it matters:

• Failing RAM = hardware. Replace it.
• Corrupted data caused by memory errors = software. Sometimes recoverable.

When faulty memory flips bits, it can damage files, the Windows registry, or system files in ways that persist after you install a new module. I've seen this twice: new RAM, clean memory tests, and yet Windows still behaving strangely because the old module had chewed through a few critical files before it died.

What does corrupted RAM look like?

Corrupted RAM—by which people usually mean the data corruption it causes—looks like unexplained software failures. Files that won't open or open garbled. Programs that install successfully then refuse to launch. Settings that reset themselves. Windows Update failing at a different percentage each time.

None of these scream "memory." They look like a messy Windows install, which is exactly why so many people reinstall the OS and end up right back where they started—with the same bad stick producing the same corruption.

How do I fix corrupted RAM?

Two steps, in this order. First, replace the faulty module—you can't repair corrupted memory any more than you can repair a cracked screen. Second, repair the damage it caused:

1. Run sfc /scannow and DISM /Online /Cleanup-Image /RestoreHealth from an elevated command prompt to rebuild damaged system files.
2. Reinstall any application that keeps misbehaving. A corrupted install rarely fixes itself.
3. Restore personal files from backup. If you don't have one, now is a genuinely good moment to start.

On my own machine, the reinstalled Windows ran flawlessly for months after the bad module was swapped out—proof that the earlier chaos was hardware, not software.

What if you can't test on your own machine?

If your PC won't POST, you can't run software tests at all. Two paths remain.

The first is a USB boot test on another computer. Pull a suspect stick, put it in a working machine, and run MemTest86 there. A word of caution: if the module is genuinely faulty, you're running your friend's machine with bad memory in it, which can corrupt their data too. Use a machine you don't mind risking, or one where you can back up first.

The second is a memory card tester—a hardware tool that slots the module and checks it without any system at all. These exist in repair shops. They're not worth buying for a one-time check, but a local technician will run a diagnostic for a modest fee, and if you're not comfortable opening the case, that's money well spent.

Replacing a bad module, and when RMA is worth it

Most RAM carries a lifetime warranty from major manufacturers. If MemTest86 flags errors and the stick is less than, say, a decade old, you may be entitled to a free replacement. Check your original receipt and the brand's RMA policy before you buy anything.

If you do buy, match the existing specs where possible—same generation (DDR4/DDR5), same speed, same capacity. Mixing wildly different modules can work, but it sometimes introduces instability that mimics a memory fault. After all this testing, the last thing you want is to create a new problem.

And once the new module is installed, run MemTest86 again. Not because you doubt the replacement, but because that's the only way to know for certain that your crashes are gone and your data is safe.

One last thing worth saying out loud

The scariest part of failing RAM isn't the blue screens. It's the silent corruption that happens before the crashes start—files quietly written wrong, backups quietly saved damaged. If you've ever had a MemTest86 run light up with hundreds of errors while your machine seemed to be working fine, you understand. The computer was lying to you the whole time.