First, before anything else: power it off and leave it at room temperature

If the drive is running now, shut it down and disconnect it. If it has already stopped, do not power it on again. Leave it at normal room temperature. Do not put it in a fridge or freezer, do not open it, and do not tap or shake it.

Every extra power-on of a mechanically damaged drive can change the case. The best move right now is the smallest one: keep the situation as it is, and write down what happened before the inspection.

The last thing that happened before the fault: a drop, a power cut, a format, or a sudden stop.
The sound on connection: repeated clicking, a whine, or complete silence with no spin-up.
Whether the drive appears in the system, and the exact error message if there is one.
What has been tried so far: recovery software, a different cable or enclosure, or a repair attempt.

Where the freezer idea actually came from

On older drives the heads came to rest on a landing zone on the platter surface itself when power was removed — a contact start-stop design that stayed common in 3.5-inch desktop drives into the mid-to-late 2000s. Over time a surface adhesion known as stiction could develop between head and platter, and the motor would fail to spin the platters on start-up. Cooling contracted the metal slightly, and that could break the adhesion for a few minutes.

There was a second, less-known reason: an electronic component or a solder joint on the edge of failure can behave differently while cold and work briefly. That is why the stories of "it worked for me" spread, and some of them were true. But it was a short window on a completely different generation of drives, and it was never a repair.

Why the trick cannot work on a modern drive

Modern drives park their heads off the platter surface on a dedicated ramp, so the heads never touch the surface at rest. That transition started with 2.5-inch mobile drives and was broadly complete in desktop drives by around the end of the 2000s, and the original cause of the myth went with it.

Storage density changed too. The head now flies at a height measured in nanometres over extremely fine tracks, and the servo system is calibrated for a known operating temperature range. Cooling a drive by tens of degrees pushes its mechanical dimensions outside the range the servo can compensate for, so read errors go up, not down.

The common failures today — a failed head, a scratched surface, corrupted firmware modules, a dead board, or a seized spindle — are not addressed by changing temperature in any way.

What the freezer actually does to the drive

A hard drive is not sealed the way it looks. Most drives have a filtered breather hole that equalises internal air pressure with the outside. When a cold drive is brought back into a warm room, humid air is drawn in and water condenses inside the platter chamber. Helium-filled drives are genuinely sealed, and cooling those adds nothing but pointless stress.

The important distinction is this: the original fault may well be recoverable in many cases, while scratches and corrosion turn part of the data into permanent loss that nobody can recover. A plastic bag does not prevent it. The drive breathes, and the cold-then-warm cycle repeats with every attempt.

Condensation: a microscopic droplet on the platter is orders of magnitude taller than the head's flying height, so the head strikes it on the first spin-up.
Corrosion: moisture attacks the thin magnetic layer and the electrical contact points, and the damage continues after the drive dries.
The lubricant layer: the platter carries a lubricant film only molecules thick, and cooling plus condensation displaces and contaminates it.
The motor: spindle bearing oil becomes more viscous when cold, so the drive may not reach speed, or the motor is stressed at start-up.
Contamination: any particle inside the chamber becomes a permanent scratch along the head's path, and a scratch erases the magnetic layer for good.

Five more myths worth stopping for

What these fixes have in common is that they all assume the cause of the fault is known before it has been inspected. In recovery, a wrong assumption costs more than waiting.

Tapping or shaking the drive: the idea is to free a stuck head. In practice the shock can push the head into contact with the surface or drag it across, turning a workable fault into a permanent scratch.
Swapping the PCB from an identical model: today's board holds calibration data unique to that specific drive, stored in memory on the board, so a matching model and firmware revision is not enough, and fitting a donor board can cost access to the data or corrupt the translation tables. If the fault arrived over the power supply, the donor board can fail the same way, which is why electrical inspection comes before any board is fitted.
Running recovery software on a clicking drive: clicking is usually a struggling head repeatedly recalibrating. Software forces continuous reads and retries on bad sectors, extending contact time and spreading debris — and installing it on the affected drive itself can overwrite exactly what you are trying to recover.
Opening the drive at home to clean it or reseat a head: a quiet room is not a clean environment, and dust you cannot see is enough to become a permanent scratch along the head's path. Refitting heads also needs tooling that holds their alignment over the surface.
Turning the drive over or running it on its side to make it spin: orientation does not repair a failed head or a seized motor, and the only certain result is another power-on cycle on an unstable drive.

What actually decides the outcome

What determines the recovery chance is not one trick. It is how many times the drive was powered on after the fault, whether the head has touched the surface, and how much new data has been written over the old.

The professional path looks nothing like the freezer: the drive is inspected first to identify the fault type, opened when needed in a clean environment with filtered airflow, and read with imaging hardware that controls head selection, timeouts and read order — then the work is done on an image, not on the original disk.

There is no home substitute for that path. But the first step is yours: keep the drive powered off and in its current state until it is inspected.

Does Putting a Hard Drive in the Freezer Fix It?