Why RAM Testing Matters
RAM problems are unusual compared to most hardware failures. A dead GPU just won't output video. A dead PSU won't power on at all. Bad RAM often works fine most of the time, then corrupts a single bit of data during one specific operation, and the effect shows up somewhere completely unrelated. A game crashes. A file gets corrupted during a save. Windows blue screens with an error code that points at a driver instead of the memory itself.
That's what makes RAM one of the hardest components to diagnose by symptoms alone. You can spend hours reinstalling drivers, rolling back Windows updates, or reseating your GPU when the actual fault is a memory module quietly flipping bits under load. A proper memory test removes the guesswork by hammering every address in your RAM with patterns designed to expose exactly this kind of fault.
Signs Your RAM Might Be the Problem
Bad RAM tends to produce a specific pattern of symptoms rather than one clear signal. Random blue screens that reference different files or drivers each time is a strong indicator, since a consistent software bug usually points to the same error repeatedly. Programs crashing unpredictably, especially ones that were stable before, is another sign worth paying attention to.
Corrupted files are one of the more concerning symptoms, because it means bad data was actually written to disk. If you notice a document or save file that opens with garbled content, or a game that keeps reporting corrupted save data, that's worth investigating with a memory test sooner rather than later. Failed installations, especially operating system installs that error out partway through for no obvious reason, also point toward memory as a likely cause.
What MemTest86 Actually Does
MemTest86 runs outside of your operating system. It boots directly from a USB drive before Windows or any other OS loads, and it takes full control of your RAM to write specific bit patterns across every address, then reads them back to check whether what comes out matches what went in. If even a single bit doesn't match, that's a hardware fault, not a software glitch, a driver issue, or a corrupted file.
Running the test outside the operating system matters because Windows itself uses RAM constantly, which means a test running inside Windows can never check every address cleanly. MemTest86 sidesteps that entirely by taking over before any OS is loaded, letting it test close to the full capacity of your installed memory.
Creating a Bootable MemTest86 USB
Download the MemTest86 image from its official site onto a working computer, and grab a spare USB drive of at least a few gigabytes. Everything on that drive will be erased during this process, so make sure it's empty or something you don't mind wiping.
MemTest86 provides its own imaging tool that writes the bootable image directly to the USB drive. Run that tool, select your USB drive from the list, and let it complete the write. Once it finishes, the drive is bootable and ready to use in the PC you actually want to test. Don't just drag and drop the downloaded files onto the USB drive, since that won't create a bootable image on its own.
Running the Test in BIOS
Plug the USB drive into the PC you want to test, then restart it and enter the boot menu. This is usually a specific key pressed right after power on, commonly one of the function keys or Delete, and it varies by motherboard brand. From the boot menu, select the USB drive instead of your normal boot drive.
MemTest86 will load its own interface directly, without touching your installed operating system. From there, you can start the default test sequence, which covers a broad range of patterns designed to catch common memory faults. Just let it run. There's no configuration required for a standard check.
Reading the Results
MemTest86 runs through multiple test passes, and each one uses a different pattern to stress the memory in a different way. As it works through the addresses, it displays a running error count. Zero errors after a complete pass is a good sign. Any errors at all mean the test found a mismatch between what was written and what was read back, which points to a hardware fault somewhere in your memory subsystem.
Pay attention to which module the errors are attributed to if you have multiple sticks installed, since MemTest86 will often indicate which physical slot or address range the fault came from. That information saves you from having to test each stick individually right away.
What to Do If Errors Show Up
If you get errors, the next step is isolating which stick is actually at fault. Power down, remove all but one RAM module, and run the test again with just that stick installed. Repeat this process one stick at a time, cycling through each module and each slot individually. This tells you whether the fault follows a specific stick, a specific slot, or shows up regardless of configuration.
If the errors follow a specific stick no matter which slot it's in, that module is faulty and needs to be replaced, usually through the manufacturer's warranty if it's still covered. If the errors only show up in a specific slot regardless of which stick is installed, the problem may be the motherboard slot itself rather than the RAM. And if reseating the module clears the issue entirely, it may have simply not been seated fully in the first place.
How Long to Run It and When to Stop
A single complete pass catches most outright hardware faults, but subtle issues sometimes only show up after multiple passes as the test cycles through different bit patterns and address orders. If you suspect intermittent instability rather than a hard failure, let it run for several hours or overnight to get through multiple full passes.
If you get through two or more complete passes with zero errors, your RAM is very likely healthy and the instability you were chasing has a different cause elsewhere in the system. That's still useful information. It rules RAM out and points you toward checking your PSU, storage, or software configuration instead of continuing to suspect memory that's actually fine.