Computer Freezes Randomly? How to Find and Fix It

Nine times out of ten when someone messages us about a computer that freezes randomly, the culprit turns out to be a driver that went sideways after an update — that’s somewhere around 40 percent of all the freezing tickets I handle in a given week. Overheating accounts for maybe a quarter, bad RAM another 15-ish percent, and then you get the grab-bag of failing hard drives, sketchy power supplies, and the occasional cryptomining malware someone picked up from a dodgy torrent site. But here’s the thing that saves you from chasing all of those at once: if you boot into Safe Mode and the freezing stops, your problem is software. Still freezes in Safe Mode? Hardware.

The 60-Second Test That Splits Everything in Two

Freeze Diagnostic

Boot into Safe Mode → use the PC for 20 min

▼

Does it still freeze?

NO — runs fine in Safe Mode

Software problem

1. Check Event Viewer timestamps

2. Roll back GPU / WiFi driver

3. Clean boot to isolate startup app

→ Usually a $0 fix

YES — freezes in Safe Mode too

Hardware problem

1. MemTest86 overnight (RAM)

2. Check GPU/CPU temps (HWiNFO)

3. CrystalDiskInfo (storage)

→ Identify failing part

Safe Mode strips Windows down to nothing — just the bare Microsoft drivers, no third-party startup junk, no GPU driver, no fancy audio stack. That’s what makes it so useful as a diagnostic, because if your machine runs clean in Safe Mode for twenty minutes and then freezes sixty seconds after a normal boot, you know for a fact something in the driver or software pile is breaking things.

Getting there on Windows 10 or 11: hold Shift, click the power icon, click Restart (keep holding Shift). Blue recovery screen shows up — Troubleshoot, Advanced Options, Startup Settings, Restart. Press 4 for plain Safe Mode, 5 if you need WiFi. Forty seconds, maybe less once you’ve done it before.

Once you’re in, just use the machine. Open stuff, move files, let it sit idle, whatever. If it runs perfectly for twenty minutes, you’ve got your answer.

About three weeks back a guy named Marcus down in Phoenix messaged us — Lenovo ThinkPad T14 Gen 4, random one-to-three-second freezes happening maybe four or five times a day for close to a month. His company’s IT department had told him to reinstall Windows. He did, which took half his Saturday. Freezing came right back within two days because Windows Update just re-downloaded the exact same problematic driver that caused it in the first place. I walked him through the Safe Mode test over WhatsApp. Twenty minutes in Safe Mode, not a single freeze. We pulled up Event Viewer and there it was — Intel Wi-Fi 6E AX211 driver crash entries perfectly timestamped to every single freeze he’d reported. Rolled the driver back to the previous version from Intel’s own site. Six minutes. Done. That $0 fix would have been a $200 repair ticket at a shop that doesn’t bother checking Event Viewer.

Software Causes (Safe Mode Doesn’t Freeze)

Drivers — Almost Always Drivers

I’m not exaggerating when I say the GPU driver is the single biggest source of random freezing on modern Windows PCs. NVIDIA ships new Game Ready updates roughly every two weeks and about once a quarter one of those introduces a regression where the display just hangs for a few seconds on certain monitor combos. AMD has the same problem with their optional driver track, a bit less often but nasty when it hits. Intel’s Arc drivers were a mess for most of 2025 — better now but still not perfect.

You can actually see this happening in real time if you know where to look. Press Win+X, open Event Viewer, click Windows Logs then System, sort by Level. What you’re looking for: “Display driver nvlddmkm stopped responding and has successfully recovered.” That’s NVIDIA’s kernel driver crashing and Windows recovering it. AMD equivalent is “amdkmdap stopped responding.” If you see Kernel-Power Event ID 41 right after one of these display errors, the GPU driver hung long enough that Windows gave up and hard-rebooted.

The proper fix — and I cannot stress this enough — is a clean driver install, not just updating on top of the old one. Download DDU (Display Driver Uninstaller, free, from wagnardsoft.com), boot into Safe Mode, run DDU, let it strip every trace of the existing driver, reboot normally, then install a fresh copy straight from nvidia.com/drivers or amd.com/en/support. Don’t let Windows Update sneak its own driver version back in — go to Settings, Windows Update, Advanced Options, Optional Updates, and decline anything GPU-related. Whole process takes maybe fifteen minutes and honestly it fixes so many problems I sometimes think NVIDIA should just bundle DDU with every driver release.

WiFi and Bluetooth drivers are the second biggest offenders. Intel’s Wi-Fi 6E AX211 and that Killer 1675 chipset have been particularly terrible in late 2025 and early 2026 — random two-second freezes that look exactly like a GPU problem until you check Event Viewer and see it’s the wireless adapter crashing instead. Device Manager, find your wireless adapter, right-click, Properties, Driver tab, Roll Back. If Roll Back is greyed out, uninstall the device entirely (tick the “delete the driver software” box) and grab the driver from Intel or Realtek’s actual website.

A Bad Windows Update

This one bites people maybe two or three times a year. Microsoft pushes a cumulative update on Patch Tuesday, something in it breaks on a subset of hardware, and suddenly you’re getting random freezes that started the exact same week the update installed itself overnight without asking. The symptoms look identical to a driver issue — random hangs, clock frozen on screen — except the timing lines up suspiciously with that update.

Here’s how to check: Settings, Windows Update, Update History. Look at dates. Did your freezing start within a day or two of the last cumulative update? If yes, uninstall it — Installed Updates, find the suspect, right-click, Uninstall, reboot. If the freezing stops, pause updates for a couple weeks until Microsoft fixes whatever they broke. Five, maybe ten minutes.

Too Many Things Running At Once

Got a message from a graphic designer in Seattle — I’ll call her Priya — with a workstation packing 64GB of RAM. Machine was freezing every two hours, almost on a schedule. She was convinced the RAM was defective because 64 gigs ought to be enough for anything, right? I had her open Task Manager and sort by Memory. Chrome with 52 tabs: 14GB. Figma with three projects: 11GB. Slack: 2.4GB. Docker with four containers: 12GB. Adobe stuff humming in the background: another 8GB. That’s 47 and change out of 64, and every time a spike pushed past what was physically available, Windows tried to swap to the page file — which someone had set up on her mechanical D: drive instead of the NVMe. Moved the page file, trimmed the Chrome tabs, no more freezes. Total cost: zero. Total time: eight minutes and a conversation about tab hoarding being an actual lifestyle problem.

The quick version for most people: Ctrl+Shift+Esc for Task Manager, click the Startup tab, disable anything you don’t actively need. Be aggressive about it — you can always re-enable something if it turns out you needed it. If Chrome is eating most of your RAM before freezes happen, see our Chrome memory usage guide — running out of RAM forces disk swapping which feels exactly like a freeze.

Malware That Eats Your CPU

Cryptomining malware pegs your CPU or GPU at 100 percent around the clock, which causes overheating, thermal throttling, micro-freezes. Dead giveaway: fans screaming at full speed while you’re doing nothing more strenuous than checking your email, or Task Manager showing 95-100 percent CPU usage with some process you’ve never heard of eating all of it.

Boot into Safe Mode, run Malwarebytes Free. Takes about thirty minutes for a full scan. If Malwarebytes comes up clean but your CPU is still maxed out, look at Task Manager for process names that sound legitimate but aren’t — “System Update Service” and “Runtime Broker” are popular disguises for miners. Windows Defender Offline Scan catches the rootkit-level stuff that Malwarebytes sometimes can’t see.

Hardware Causes (Freezes Even in Safe Mode)

Bad RAM — Hardest to Pin Down

This is my least favorite diagnosis to make because there’s no pattern to find. The machine freezes during gaming, freezes at idle, freezes while you’re browsing Reddit at 2 AM. Sometimes you get a blue screen with MEMORY_MANAGEMENT or IRQL_NOT_LESS_OR_EQUAL. More often the screen just locks up completely — no error, no crash dump, nothing — and you end up holding the power button for five seconds.

The built-in Windows Memory Diagnostic (mdsched.exe) gives people a false sense of security here. It runs two quick passes with basic test patterns, takes fifteen minutes, and says everything’s fine. The problem is that marginal RAM — a stick that flips one bit out of every ten thousand reads — passes that test easily because the test isn’t aggressive enough to trigger the fault. MemTest86 from memtest86.com is what you actually want. Boot it from a USB stick, let it run overnight — minimum four passes, ideally eight. That’s how you catch the intermittent stuff.

I had a customer named Andre in Atlanta not long ago — brand new gaming build, i7-14700K, RTX 4070 Ti, 2x16GB G.Skill Trident Z5 DDR5. Random hard locks about once a day, sometimes mid-game, sometimes just sitting on the desktop with Steam open. He’d already run Windows Memory Diagnostic (passed), checked his drives with CrystalDiskInfo (healthy), reinstalled Windows from scratch (still froze). He was one step away from RMA’ing his ASUS motherboard when I told him to run MemTest86 overnight. Passes one through three came back clean. Pass four: twelve errors. One of the two sticks had come defective from the factory — which happens with DDR5 more than you’d think because the tolerances on those things are genuinely tight. Pulled the bad stick, ran single-channel on the good one, rock stable. G.Skill sent a warranty replacement in four days, no questions asked. That’s a $0 fix if you catch it yourself, versus a $200+ bill if some shop replaces your entire motherboard chasing a problem that was never the board.

Overheating

Different story depending on whether you’ve got a desktop or a laptop. Desktops overheat because thermal paste dries out after three to five years and nobody replaces it, or because a case fan failed silently six months ago and you never noticed. Laptops overheat because the heat pipe fins pack solid with dust after twelve to eighteen months of daily use, especially in thin machines — those 13 and 14-inch ultrabooks with the pretty chassis have basically zero thermal margin once the fins are 30 percent clogged.

Grab HWMonitor from cpuid.com and watch your CPU temperature while you do something heavy — game, render, pile up Chrome tabs, whatever makes your fans spin. If your CPU breaks 95°C or your GPU passes 92°C, that’s your problem. Desktop: pop the side panel, blast the heatsink with compressed air, think about fresh thermal paste if temps stay high (Arctic MX-6, eight bucks, you don’t need anything fancier). Laptop: YouTube “[your laptop model] disassembly” for the screw locations, pop the bottom panel, clean out the fan assembly. Most laptops drop 15 to 20 degrees Celsius after a proper cleaning — the difference between freezing every hour and never freezing again. We wrote a full laptop overheating guide that covers the entire repaste process if dust alone doesn’t fix it.

"If your freezes correlate with how long the machine has been running — fine for the first hour, then unreliable after that — check temperatures before chasing anything else. I've watched customers spend hundreds on RAM and SSD replacements before the actual fix turned out to be twenty dollars of thermal paste and a can of compressed air."

Dying Hard Drive or SSD

Failing storage doesn’t always announce itself with clicking noises and missing files. Sometimes the only symptom is these weird half-second freezes when the OS tries to read a bad sector or when the SSD controller runs into a problematic NAND block and has to retry the read three or four times before it succeeds.

CrystalDiskInfo — free, takes thirty seconds to install. Open it, look at Health Status. “Good” is good. “Caution” or “Bad” means back up your data right now and start shopping for a replacement drive. For mechanical drives pay attention to Reallocated Sectors Count (anything above zero is bad news) and Current Pending Sector Count. For SSDs check life remaining percentage — below 10 percent and you’re living on borrowed time. A 1TB NVMe in 2026 runs $60 to $80. Cloning your old drive over with Samsung Data Migration or Macrium Reflect takes two to three hours but it permanently solves the problem and you don’t have to reinstall anything.

Power Supply Going Out (Desktop Only)

This one has a very specific symptom pattern: freezes or hard shutdowns during heavy load — gaming, video rendering, stress testing — but perfectly fine at idle. The machine boots up, you browse the web, everything’s stable, you launch a game and thirty minutes in the whole thing locks up or just cuts to black.

PSU capacitors wear out after four to six years regardless of how nice the unit was when it was new. If yours is that old and you’re getting load-only freezes, swap in a known-good PSU before you spend hours chasing other causes. A decent 650W unit runs $60 to $100 and you’ll get another five to seven years out of it. If the freezes escalate to the point where the machine won’t turn on at all, the PSU is almost certainly toast — do the paperclip test.

The Event Viewer Shortcut

Every crash, freeze, and driver hiccup gets logged in Event Viewer with a timestamp. Most people never open it. If you learn to read just five or six Event IDs, you can pinpoint most freeze causes in under a minute.

Win+X, Event Viewer, Windows Logs, System. Sort by date and look for Error and Critical entries around the time your last freeze happened.

Kernel-Power Event 41 is the big one — means an unexpected shutdown, usually power loss or a hard hardware freeze. If it shows up alone with no preceding errors, think PSU or physical hardware. If you see Display driver events (ID 14 or 4101, nvlddmkm or amdkmdap) right before the Kernel-Power 41, that’s your GPU driver crashing. Disk errors (Event IDs 7, 11, 51, 153) point at storage failing. WHEA-Logger Event 17 means a hardware error at the CPU, RAM, or PCIe bus level — that one’s always worth taking seriously. DistributedCOM 10016 errors show up constantly and are almost always harmless — ignore them unless they line up perfectly with your freeze timestamps.

The trick: if you see the same Event ID repeating at every single freeze, you’ve found the thing.

Laptop-Specific Issues

A few freeze causes that only happen on laptops, mostly because of how cramped the internals are.

NVIDIA Optimus switching — the system tries to swap between integrated graphics and the discrete GPU when you open a game or video editor, and sometimes that handoff just fails and the screen freezes for one to three seconds. If this is happening to you, go into NVIDIA Control Panel, Manage 3D Settings, Preferred Graphics Processor, and lock it to either “Integrated” (for battery) or “High-performance NVIDIA processor” (for performance) instead of letting it auto-switch. That eliminates the handoff entirely.

Battery health matters more than people think. A battery at 40 percent or less of its design capacity can’t deliver peak voltage when the CPU spikes, which causes a momentary freeze or forced throttle. Check yours: open PowerShell as admin, type powercfg /batteryreport, look at Design Capacity versus Full Charge Capacity. If full charge is below 60 percent of design, that battery is due for replacement. If your laptop’s runtime has dropped along with the freezing, our battery drain guide covers the background-app audit that often catches the same culprits.

Thunderbolt dock conflicts are the other one — CalDigit TS4 and Lenovo’s own ThinkPad docks have been particularly bad about this. Update Thunderbolt firmware from your laptop manufacturer’s support page, not from Intel directly, because manufacturers customize the firmware for their specific hardware.

When DIY Stops Making Sense

If you’ve burned more than two hours on this without narrowing it down, more time isn’t going to change that — you’re in diminishing returns territory. We’ve handled over 400 freezing cases this year and the average remote diagnosis takes about twelve minutes. The fix usually takes another eight to fifteen minutes on top of that.

Remote PC diagnosis is $9.90. You message us on WhatsApp, we connect remotely, and most people are back to stable inside thirty minutes. Compared to losing half a Saturday to a diagnostic rabbit hole, $9.90 is math that basically sells itself.

If your freezing PC also has blue screen errors, the BSOD error code usually points directly at the failing component. If freezes happen specifically during shutdown attempts, see our Windows 11 won’t shut down guide — same hung-process pattern. And if you notice screen flickering alongside the freezes, that’s almost always a GPU driver issue — the DDU clean install process above will fix both.

Last verified: May 2026 against Windows 10 22H2, Windows 11 24H2, and current-generation hardware from Intel, AMD, and major laptop OEMs.