First: which of these is actually happening?

Be precise about the symptom. It narrows the cause enormously.

SymptomWhat it meansUsual suspects
Total freeze — mouse won't move, Caps Lock light doesn't respond, only a power-button hold worksThe kernel itself is stuckHardware: heat, memory, drive, power
Soft freeze — cursor moves, everything else unresponsive, recovers after 10–60 secondsThe system is waiting on somethingDrive I/O, memory exhaustion, a hung network mount
One app freezes — the rest of the system is fineThat program is stuckThe app, its extensions, or a file it's opening
Sudden restart or blue screenA crash, not a freezeDriver or hardware fault — the log will name it
Always slow, never stuckA bottleneck, not a faultSee our slow computer guide

If it's only one application, stop here and deal with that application — update it, disable its plugins, test with a different file. Nothing below applies.

Open the log — it names the culprit

Windows: Reliability Monitor

This is the single most useful and least known tool in Windows. Press the Start button, type reliability, and open View reliability history.

You get a timeline chart with one column per day and a red X for each failure. Click any day to see what happened, in plain language: which application stopped responding, which driver failed, which update installed. Critically, it puts everything on one timeline — so you can see that the freezes started the same day a graphics driver updated, which is the kind of connection that's impossible to make from memory.

What to look for: a repeating entry across multiple incidents. One-off application errors are noise. The same driver or the same hardware event appearing every time is your answer.

For more detail, Event Viewer (eventvwr) → Windows Logs → System, filtered to Critical and Error. Two entries are worth knowing:

macOS: Console and crash reports

Open Console (Applications → Utilities). In the sidebar, Crash Reports lists individual application failures, and Log Reports holds system-wide records. A full system freeze that ended in a restart leaves a panic report — look for a file beginning with "Kernel" or "panic" dated to your incident.

You don't need to interpret the whole report. Look at the first few lines for a named extension or driver — anything ending in .kext, or a third-party product name. That name is usually the whole answer.

The pattern tells you the cause

While you're gathering evidence, note when it freezes. This is often more diagnostic than the logs:

Heat

The clearest signature: fine from cold, degrades or freezes under sustained load, fans loud, chassis hot. Modern processors slow themselves down first and shut off as a last resort, so a hard freeze from heat usually means the cooling has degraded significantly.

Install a temperature monitor and watch while you work. Sustained temperatures near the manufacturer's maximum under normal use point at cooling. What to do:

A failing drive

This is the most urgent possibility, because unlike the others it takes your data with it. Signature symptoms: soft freezes of 10–60 seconds that recover on their own, freezes when opening large files, file operations that hang, occasional corrupted files, or a machine that takes far longer to boot than it used to.

Back up before you diagnose

If any of the above describes your machine, copy your important files to an external drive or cloud storage now, before running tests. Diagnostics put a failing drive under load, and load is exactly what finishes them off. A drive that is throwing errors can go from "mostly working" to "unreadable" in a single session.

Then check the drive's own health data:

Any non-zero and rising count of reallocated or pending sectors means replace the drive. Not "monitor it" — replace it. Drives rarely recover.

Bad memory

Faulty RAM produces the most maddening pattern: random freezes and crashes with no relationship to what you were doing, sometimes with corrupted files thrown in. It's also easy to test definitively.

On a desktop or an older laptop with accessible slots, there's a free physical test too: if you have two or more sticks, run with one at a time and see whether the freezes follow a particular stick or a particular slot. Also simply reseat the modules — memory can work its way loose over years of heating and cooling, and reseating fixes a genuine share of "random freeze" machines at no cost.

Drivers, especially graphics

If Reliability Monitor names a driver, or freezes only happen in games and video work, start here.

The test that separates hardware from software

If you've done the above and still don't know, this settles it. Boot into a minimal environment and use the machine the way that normally triggers a freeze.

Freezes stop in Safe Mode → software: a driver, a startup program, or a background service. Re-enable startup items in batches to find it. Freezes continue in Safe Mode → hardware, and the memory and drive tests above are where to concentrate.

The short version

Name the symptom precisely — total freeze, soft freeze, one app, or crash. Open Reliability Monitor (Windows) or Console (Mac) and look for a name that repeats across incidents. Note when it happens: under load means heat, on file operations means the drive, at random means memory. If anything points at the drive, back up first, then test. If you're still stuck, Safe Mode tells you whether it's hardware or software in twenty minutes.