How to Check If RAM Is Working Properly

Faulty memory is one of the most frustrating faults in a PC because the symptoms never look like a memory problem. You get a blue screen one day, a corrupted file the next, and a game crash the day after, with nothing obviously connecting them. Knowing how to check if RAM is working properly turns that guessing game into a clear yes or no.

The good news is that RAM is one of the easiest parts to test with certainty. The tools are free, the tests are pass or fail, and if a stick is bad you can usually prove exactly which one it is without spending anything.

To check if your RAM is working, first open Task Manager (Ctrl + Shift + Esc), select the Performance tab, and click Memory to confirm Windows sees the full installed amount at the right speed. Then run Windows Memory Diagnostic for a quick test, and boot MemTest86 from a USB stick for a thorough one. Healthy RAM completes a full test run with zero errors.

1. Confirm Windows Sees All Your RAM

Before running any test, check that all the memory you installed is actually being detected. Press Ctrl + Shift + Esc to open Task Manager, go to the Performance tab, and click Memory.

• Total amount: should match what is physically installed. Two 8 GB sticks should show 16 GB, minus a small slice of hardware-reserved memory.
• Speed: shown in MT/s or MHz. If a 3200 kit reports 2133, the memory works but is running at its slower default, which is covered further down.
• Slots used: "2 of 4" should match the number of sticks you actually fitted. A stick Windows cannot see will not show up here.

A quick way to see the same information in one place is the Overview tab of our free PC Reporter, which also reports whether the sticks are running in dual channel mode. Two sticks running in single channel still pass every health test, but they give away real performance for free.

2. What Startup Beeps and the DRAM Light Tell You

When memory has failed completely, the machine usually will not reach Windows at all. During POST, the motherboard checks that the memory responds, and it reports a failure through beeps or diagnostic LEDs.

Beep patterns differ between BIOS makers, so always confirm against your motherboard manual, but repeated short beeps at power-on frequently point at memory. Many modern boards skip beeps entirely and use a row of small LEDs labelled CPU, DRAM, VGA, and BOOT instead. A DRAM light that stays lit means the board stopped at the memory check; we cover what the DRAM light means and the reseating routine that usually clears it in the RAM cleaning guide.

It is also worth a minute in the BIOS itself. The main screen shows the total memory detected, and many boards list each slot individually, which tells you straight away if one stick has gone missing. If you have not poked around in there before, our guide on what the BIOS does at startup shows what to expect.

3. Symptoms of Faulty RAM

Failing memory corrupts data in whatever happens to be loaded at that address, so the symptoms move around. These are the classic signs:

• Blue screens with memory-related codes: MEMORY_MANAGEMENT, PAGE_FAULT_IN_NONPAGED_AREA, or IRQL_NOT_LESS_OR_EQUAL appearing at random times.
• Crashes that move between programs: the browser today, a game tomorrow, with no single app to blame.
• Corrupted files and failed installs: archives that will not extract, Windows updates that fail repeatedly, or documents that suddenly will not open.
• Freezes under load: the machine locks up when many things are open, then behaves when lightly used.
• Sudden restarts with nothing on the screen: no error, just a reboot.

None of these prove the memory is bad on their own. An unstable processor or failing power supply produces near-identical behaviour, which is why the companion guide on how to check if a CPU is working properly starts from the same symptom list. The difference is that memory can be tested directly and conclusively, so it is the right place to start.

4. Run Windows Memory Diagnostic

Windows ships with a basic memory tester, and it is the fastest way to a first answer. Press Windows + R, type mdsched.exe, press Enter, and choose Restart now and check for problems. Save your work first, because the test runs outside Windows during a reboot.

1. Let the Standard test run: it starts automatically and takes roughly 10 to 30 minutes depending on how much memory you have.
2. Optional, press F1 for options: the Extended test is much slower but more thorough, and you can raise the pass count for a longer soak.
3. Read the result after the reboot: a notification appears shortly after you log back in. If you miss it, open Event Viewer (Windows + X), go to Windows Logs, then System, and look for entries from MemoryDiagnostics-Results.

5. Run MemTest86 for a Definitive Answer

MemTest86 is the tool the rest of the industry measures against. It boots from a USB stick before any operating system loads, which lets it walk through the whole of memory with a series of test patterns designed to trip up weak cells.

How to run it

1. Make the boot stick: download the free edition from the official MemTest86 site and write it to a spare USB drive with the imaging tool included in the download. Anything already on the drive is erased.
2. Boot from the USB drive: restart, tap the boot-menu key for your board (often F8, F11, or F12, shown on the startup screen), and pick the USB stick.
3. Let it run to completion: the default configuration runs four full passes. On a large kit that can take several hours, so overnight is the practical choice.
4. Read the summary: the result is a plain pass or fail, and any errors are listed in red with the memory address where they occurred.

Memory testing tools compared

6. Find the Bad Stick

A failed test tells you the memory system has a fault, not which part of it. With more than one stick installed, narrow it down by testing sticks individually.

1. Test one stick at a time: remove all but one stick, run MemTest86, and repeat for each stick in the same slot. The stick that produces errors alone is your culprit.
2. Then swap slots: put a known-good stick into the slot the bad one was using. If the errors follow the slot rather than the stick, the problem is the motherboard slot, not the memory.
3. Clean and reseat before you conclude: oxidised or dirty contacts produce errors that look exactly like a dying stick. The full routine is in our guide on how to clean RAM contacts, and it has rescued plenty of "faulty" memory on my bench.

On laptops the same logic applies where the memory is accessible. Some machines pair one soldered bank with one removable stick, and some are fully soldered. Testing still gives you the verdict either way; the difference is that a soldered failure means a board repair or a warranty claim rather than a cheap stick swap.

7. When RAM Only Fails at Its Rated Speed

Memory advertised at higher speeds does not run there by default. The sticks boot at a conservative standard speed, and the advertised figure is stored as an XMP profile you switch on in the BIOS.

This matters for testing because that profile is, formally, an overclock. A kit can be error-free at its default speed and throw MemTest86 errors with the profile enabled, especially with four sticks fitted or on an older CPU's memory controller. That combination does not necessarily mean the sticks are defective.

• Errors with XMP on, clean at stock: the sticks are healthy but the profile is unstable on your particular board and CPU. Try the profile one speed step lower, or leave it off.
• Errors at stock speed: a genuine fault. Stock settings are the baseline the memory must pass.

Mixed kits deserve a mention here, because two sets of sticks bought separately can each pass alone yet refuse to run stably together at the rated profile. Our RAM configuration optimizer shows which combinations of sticks and slots make sense for your board before you buy.

8. Common Problems and Solutions

Windows Shows Less RAM Than Installed

A small amount listed as "hardware reserved" is normal; the system and integrated graphics claim a slice for themselves. A whole missing stick is different. Power down, reseat the stick until both clips lock, clean the contacts if it has been in place for years, and check the BIOS per-slot readout to see whether the board detects it at all. If the stick refuses to appear in any slot, it has likely failed outright.

New RAM Causes Crashes or Will Not Boot

First, give it time: modern DDR5 boards can sit on a blank screen for a few minutes on the first boot while they train the new memory, and interrupting that looks exactly like a dead kit. If it still will not boot, try one stick alone, confirm the sticks are in the slots your motherboard manual recommends, and check the board supports the kit's capacity and speed. Mixing the new sticks with the old ones is the most common cause of all; test the new kit on its own before blaming it.

Errors Come Back After a Clean Test

A stick can be marginal rather than dead, producing errors only when warm or under sustained load. Run MemTest86 overnight rather than for a single quick pass, and retest with XMP off to separate a weak stick from an unstable profile. If long stock-speed runs stay clean but crashes continue, the fault is likely elsewhere, and the CPU checks are the logical next step.

Repair or Replace?