A CPU that runs fine at idle but spikes to 90°C or higher the moment you start gaming, rendering, or compiling code is one of the most common — and most fixable — PC problems. Left unaddressed, sustained overheating triggers thermal throttling (reduced performance) and, in extreme or prolonged cases, can shorten the lifespan of the CPU and surrounding components. This guide works through the causes in order of likelihood, from the quick checks to the more involved fixes.

Know your limits: Most modern Intel and AMD CPUs are rated to run safely up to 90-100°C under sustained load without damage, since built-in thermal protection will throttle or shut down the system before failure. The goal of this guide isn't just to avoid damage — it's to get temperatures into a healthy range (typically under 80-85°C under sustained load) so your CPU isn't throttling and losing performance.

Step 1: Confirm what you're actually seeing

Before changing anything, get accurate data with HWMonitor or HWiNFO64 running in the background during a real workload (a game, a render, or Cinebench R23's 10-minute stress test).

  • Note the package temperature and, if shown, individual core temperatures — a large gap between cores can point to uneven cooler contact.
  • Check whether the CPU is throttling: HWiNFO64 shows a "Thermal Throttling" or "PROCHOT" flag when the CPU is actively reducing clock speed to manage heat.
  • Compare against your specific CPU model's rated maximum (TjMax), listed on Intel Ark or AMD's product pages — this varies by model, so don't assume a generic number.

Step 2: Check the obvious — airflow and dust

This is the most common cause and the easiest to rule out.

  1. Open the case and visually inspect the CPU cooler's fins and fan for dust build-up. Even a moderate layer significantly reduces heat dissipation.
  2. Check that case fans are actually spinning, correctly oriented (intake fans pulling air in, exhaust pushing it out), and that none have failed silently.
  3. Confirm nothing is blocking airflow — cables draped across the cooler, the case positioned against a wall with no vent clearance, or a case in an enclosed desk cupboard with poor ventilation.
  4. If you've recently cleaned the PC, check that fan cables weren't accidentally left unplugged during reassembly.

Give the cooler a thorough clean with compressed air (see our guide on deep-cleaning dust from your PC case) and retest before moving further down this list.

Step 3: Check the CPU cooler is actually mounted correctly

A poorly seated cooler is the second most common cause and is easy to miss because the PC will still boot and run — just hot.

  • Air coolers: Check that all mounting screws are tightened evenly, and that the backplate on the rear of the motherboard is sitting flush, not tilted. An uneven mount creates a gap on one side of the CPU that paste alone can't bridge.
  • AIO liquid coolers: Confirm the pump block is fully seated and screws are tightened in a diagonal pattern (like a car wheel) rather than one side fully tightened first, which can leave the block sitting at a slight angle.
  • Plastic backplate stock coolers (common on budget/OEM builds): These push-pin coolers are notorious for uneven pressure — if you're using the original stock cooler and pins, check each of the four pins has fully clicked through the backplate.

Remove the cooler entirely and inspect the thermal paste pattern on both the CPU lid and cooler base. An even, thin spread across the whole surface indicates good contact; a paste pattern concentrated on one side or with clear gaps points directly to a mounting problem.

Step 4: Reapply thermal paste

Thermal paste dries out over time (typically 2-4 years for standard non-metal pastes) and a bad or old application is a frequent, invisible cause of high temperatures.

  1. Power down, unplug, and remove the cooler following the steps above.
  2. Clean off all old paste from both the CPU integrated heat spreader (IHS) and the cooler's base using isopropyl alcohol (90%+) and a lint-free cloth or coffee filter — cotton wool can leave fibres behind.
  3. Apply a fresh pea-sized dot (roughly 5mm) in the centre of the CPU lid for most consumer CPUs, or a thin line down the centre for larger dies (Threadripper, some high-core-count chips) — check the paste manufacturer's guidance for your specific CPU.
  4. Reseat the cooler, tightening screws evenly and diagonally, and reconnect the fan header to the correct motherboard header (usually labelled CPU_FAN).

Step 5: Check BIOS power and fan curve settings

Sometimes the hardware is fine but the settings are working against you.

  • Fan curves: In the BIOS/UEFI (usually accessed by pressing Delete or F2 during boot), check the fan control section. If fans are set to a very quiet/silent profile, they may not be ramping up fast enough under sudden load spikes. Set the CPU fan curve to "Standard" or a custom curve that ramps more aggressively above 60-70°C.
  • Power limits (Intel): Newer Intel CPUs can draw significantly more power than their rated TDP under default motherboard settings (a practice called "unlocking" PL1/PL2 limits), causing much higher temperatures than the CPU's stock rating would suggest. If temperatures seem excessive for the CPU model, check whether your motherboard is applying an aggressive out-of-the-box power profile and consider setting PL1/PL2 to Intel's default specification.
  • PBO (AMD Precision Boost Overdrive): Similarly, AMD motherboards sometimes enable PBO by default, pushing voltage and clocks higher than stock. Disabling or tuning PBO can meaningfully reduce peak temperatures at a small performance cost.
  • Undervolting: If you're comfortable with more advanced tuning, a modest CPU undervolt (via BIOS or Intel XTU/AMD Ryzen Master) can lower temperatures noticeably with little to no performance loss, since much of a modern CPU's heat comes from voltage rather than clock speed alone.

Step 6: Consider whether the cooler is simply undersized

If you've addressed dust, mounting, paste, and BIOS settings and temperatures are still high, the cooler itself may not be rated for your CPU's power draw. This is common when a budget stock cooler is paired with a higher-end CPU (for example, after a mid-range motherboard upgrade path), or the case has genuinely poor airflow for any cooler to work with. Compare your cooler's specifications against your CPU's TDP/PPT rating — if there's a significant mismatch, an upgrade to a higher-capacity air cooler or AIO is the realistic fix (see our guide comparing air and liquid CPU cooling).

When to call a professional

If temperatures remain dangerously high (consistently above 95-100°C, triggering shutdowns) after working through every step above, or if you suspect the CPU itself may have been damaged by prolonged excessive heat, it's worth having a technician diagnose the system. Persistent overheating despite correct cooling can occasionally point to a failing motherboard VRM, a case with a fundamentally unworkable airflow design, or in rare cases a damaged CPU socket — issues that go beyond straightforward maintenance.

Was this guide helpful? Explore more Cooling & Cleaning guides, or browse all Desktop PC Repair articles.