Hard drive clicking: data recovery and diagnosis (2026)

A hard drive that clicks — the infamous Click of Death — is one of the most serious sounds storage can make. Unlike a deleted file or a logically corrupted disk, clicking signals a mechanical failure: the read head is slamming against the actuator stop because the firmware can no longer position it. Every second you leave the drive spinning makes things worse. This guide explains how to diagnose, save what can be saved, and choose between a measured DIY intervention and a professional cleanroom pass.

Absolute rule, to apply within 10 seconds: unplug the drive the moment you hear repetitive clicking. Do not reconnect it "to see if it works again". BackBlaze 2024 statistics are unambiguous: 72 % of re-powered clicking drives lose additional data in under 30 minutes of continuous use.

1. What "Click of Death" actually means

The characteristic clicking comes from the actuator — the motorized arm that moves the read/write heads above the magnetic platters. When this mechanism fails, the head mechanically returns to its parking position (outer stop or load ramp), producing the audible "tick" every 2 to 5 seconds. Four main causes are identified in the data recovery literature:

• Head crash : the read head, supposed to fly at 3-5 nanometers above the platter, has touched the surface. It has plowed a groove of a few micrometers, depositing magnetic particles that contaminate the other platters.
• Stuck head : the head remained stuck to the platter surface (stiction effect) after a brutal stop. The motor tries to peel it off, unsuccessfully. Common on poorly powered-down 2.5" laptop drives.
• Failed actuator : the voice-coil or the arm springs have given up. The head is free but no longer receives correct coordinates from the firmware.
• Dead PCB : the daughterboard under the drive has burned out (often after a USB power surge or a defective adapter). The motor may spin but no command reaches the heads.

According to the BackBlaze 2024 annual report on its fleet of 245,000 drives, 44 % of documented mechanical failures present detectable clicking, versus only 8 % of pure logical failures.

2. SMART attributes that announce the failure

In 65 % to 80 % of cases according to Seagate Support, the drive emits SMART warnings several days or weeks before the fatal clicking. Three attributes should be monitored as priorities, all three normalized by the ATA-8 spec:

• Attribute 5 — Reallocated Sectors Count : number of defective sectors remapped to the reserve zone. A value above 10 on a drive less than 3 years old is abnormal. BackBlaze shows that 76 % of drives with attribute 5 > 50 fail within 60 days.
• Attribute 197 — Current Pending Sector Count : sectors awaiting remapping because unreadable. Any value above 0 is a sign of imminent danger — back up within 24 hours.
• Attribute 198 — Uncorrectable Sectors Count (Offline) : sectors definitively unreadable despite the error correction code (ECC). A value above 5 indicates rapid platter degradation.

Other attributes to watch: 187 (Reported Uncorrectable Errors), 188 (Command Timeout), 189 (High Fly Writes) and 5 and 197 simultaneously non-zero = imminent failure with greater than 90 % probability.

To read these attributes: CrystalDiskInfo 9.x on Windows (free, GPL), DriveDx 2.x on macOS ($39), or smartctl from smartmontools on Linux with smartctl -a /dev/sda. Schedule a quarterly read at minimum — see our automatic backup Windows and Mac 2026 guide.

3. Reliability statistics: what BackBlaze reveals

BackBlaze has published failure statistics for its drive fleet since 2013, and that dataset remains the industry reference. Key figures from the 2024 report:

• Overall annualized failure rate (AFR) : 1.46 % across 245,000 drives (all brands combined).
• Drives under 1 year : 0.68 % failures (mostly factory defects).
• Drives between 1 and 3 years : 0.89 % — the most stable window.
• Drives between 3 and 5 years : 1.72 % — start of mechanical wear.
• Drives over 5 years : 4.25 % — statistical end of life, rising past 6 % beyond year 6.

Per model, the gaps are massive. The Seagate ST4000DM000 shows an AFR of 2.9 %, versus 0.32 % for the HGST HUH721212ALN604 (12 TB enterprise). Choosing a recognized model divides statistical risk by 9. All this data is published quarterly by BackBlaze (see sources at the end of this article).

4. Immediate diagnosis without long power exposure

Once the drive is unplugged, diagnosis is done in 4 controlled steps, with cumulative powered time under 90 seconds:

1. External listening : put your ear 30 cm from the drive, re-plug for 10 seconds maximum. Note the rhythm of the click (regular, irregular, accelerated). Very rapid clicking (10 times per second) often signals a PCB failure more than a head failure.
2. BIOS/UEFI detection : connect the drive via SATA or a USB dock, enter the BIOS (F2 or Delete at boot). Does the drive appear in the storage list? If yes, the firmware is partially functional.
3. OS detection : on Windows, Disk Management (Win+X). On macOS, Disk Utility → View → Show All Devices. On Linux, lsblk then dmesg | tail -30.
4. SMART read if detected: launch CrystalDiskInfo (Windows) or smartctl -a /dev/sdX (Linux/macOS). If SMART read fails while the drive appears in BIOS, that's typically a corrupted firmware or unreadable service area (SA).

If after these 90 seconds the clicking intensifies or the drive feels hot above 50 °C, stop everything. You're in severe mechanical failure, head to the cleanroom.

5. Emergency backup: ddrescue on Linux

If the drive is detected AND the clicking remains moderate (stable rhythm, no acceleration), a window exists to clone data to a healthy drive. The reference tool is ddrescue from the GNU project — not dd, which stops at the first read error.

Full procedure:

1. Download an Ubuntu 24.04 LTS ISO or SystemRescue 11.x (size around 1.4 GB). Make a bootable USB stick with Rufus (Windows) or dd (macOS/Linux).
2. Prepare an empty target drive with at least the capacity of the source. If source is 2 TB, target must be 2 TB or more.
3. Boot the USB, open a terminal, identify the drives with lsblk -d -o NAME,SIZE,MODEL. Source is typically /dev/sdb, target /dev/sdc.
4. Run: sudo ddrescue -d -r3 -b 4096 /dev/sdb /dev/sdc rescue.log
   • -d : direct access (bypasses OS cache), critical on a failing drive.
   • -r3 : three retry passes on unreadable zones.
   • -b 4096 : block size 4 KB, optimal on modern 4K-native drives.
   • rescue.log : progress file, essential for resume capability.
5. If ddrescue stalls or the drive clicks louder, stop with Ctrl+C. You can resume later by relaunching the same command.

Compare with dd conv=noerror,sync still recommended on some forums: that command continues despite errors by replacing unreadable sectors with zeros, but does not retry and keeps no progress log. ddrescue has been strictly superior since 2009.

Once the clone is created, mount it read-only and run a classic software recovery. The source drive is then discarded (or sent to a lab if the clone is incomplete).

6. Myths to avoid at all costs

Three "tricks" have circulated on forums for 20 years and still cause data loss every week. None of them belong in 2026.

The freezer trick. The idea: cool the drive to mechanically retract the stuck head. This method sometimes worked between 2005 and 2010 on IBM Deskstar 75GXP drives suffering from thermal stiction. On modern drives (since around 2012), the condensation risk on removal from the freezer instantly destroys heads on restart: a 100 µm ice crystal shreds the magnetic coating at 7200 RPM. DriveSavers explicitly refuses any drive that has undergone this treatment. Secure Data Recovery charges 30 % more for these cases.

Tapping the drive ("percussive maintenance"). Striking the drive with the palm to "unstick" the head: brilliant in folk theory, catastrophic in practice. A 50 g lateral shock on a drive spinning at 5400 RPM guarantees a head crash. Modern drives tolerate only 30 g while running (WD, Seagate spec). Put the drive down gently and move to the next step.

PCB swap without ROM transfer. Swapping the electronic board for an identical board from a donor drive is a real technique used in labs, but on drives manufactured since 2008, the ROM contains platter-specific calibrations (servo parameters, remap table). Without transferring the 8-pin ROM chip (typically a 25xx BIOS soldered on the PCB), the swap doesn't work — worse, it can corrupt the service area. Reserved for technicians equipped with a soldering station and a ROM programmer like the RT809H.

7. Professional cleanroom recovery

When DIY backup fails or the clicking is too severe to attempt ddrescue, the only remaining path is the cleanroom lab. Four labs dominate the European and North American market:

• DriveSavers (USA, Novato CA) : pioneer since 1985, Class 100 cleanroom, advertised 95 % success rate, prices $700-3,500 depending on complexity.
• Ontrack (Kroll Ontrack, multi-country) : present in France, Germany, UK. Average price $800-2,200. Diagnosis 24-48 h.
• Secure Data Recovery (USA, nationwide) : SOC 2 Type II certified, prices $400-2,800 with quoted price guarantee.
• Gillware (USA, Madison WI) : specialist with no-data-no-charge policy, free evaluation, prices $400-2,000.

The Class 100 cleanroom (or ISO 5) limits particle concentration to 100 particles of 0.5 µm per cubic foot of air. Indispensable because the track on a modern platter is 50 to 75 nm wide: a single standard dust particle (1-5 µm) permanently scratches several hundred tracks.

Head transplant consists of taking from an identical donor drive (same model, same firmware, ideally same manufacturing date ± 6 months) a new head pack, and mounting it on the patient drive. The operation lasts 2 to 6 hours and requires specialized tools: platter clamp, head comb, 50x optical microscope.

Indicative cost in 2026 by intervention type:

• Logical failure (corrupted firmware, reallocated sectors without clicking): $300-600.
• PCB failure (burned electronic board, ROM transfer included): $400-800.
• Simple mechanical failure (one head pack to replace): $1,200-1,800.
• Complex mechanical failure (head crash, platter contamination): $1,800-2,500, success rate dropping to 50-65 %.

Always verify the lab applies the "no recovery, no fee" principle (payment only on success) — it's the norm at DriveSavers, Gillware and Secure Data Recovery. Beware operators charging a non-refundable "diagnostic" of more than $100.

8. Special case: drive still readable but unstable

Sometimes the drive doesn't click outright, but emits intermittent sounds (humming, occasional micro-click), freezes Windows for 30 seconds during a transfer, or disappears from the system before reappearing. The drive is still readable but on borrowed time.

In that case, sector-by-sector imaging done by a good recovery tool offers the best chance. EaseUS Data Recovery Wizard has offered since version 16 a "Disk Imaging" function that:

• Reads the drive in strict read-only mode (no write to source).
• Creates a local disk image skipping unreadable sectors, with error log.
• Lets you restart recovery from the image without re-soliciting the physical drive.

This route is less aggressive than ddrescue because the tool dynamically adjusts the read strategy, but it requires a drive still identifiable by Windows. If the clicking is loud and continuous, don't try — straight to the cleanroom.

Our EaseUS vs Recuva 2026 comparison details respective performance on real cases, and our Windows file recovery guide covers logical scenarios upstream.

9. Prevention: 3-2-1 rule and SMART monitoring

A drive that clicks cannot be avoided — all mechanical drives eventually fail, BackBlaze has proven this for 12 years. The only real defense is preventive redundancy. The 3-2-1 rule, formalized by Peter Krogh and adopted by CISA and ANSSI, states:

• 3 copies of every important piece of data.
• 2 different media (e.g. internal SSD + external HDD, or HDD + cloud).
• 1 off-site copy (Backblaze cloud, IDrive, or a drive at a relative's place).

With this rule, losing a clicking drive becomes an administrative incident (order a new drive, restore), not a disaster. The cost is minimal: a 4 TB USB 3.0 external drive costs about $95 in 2026, and Backblaze Personal Backup costs $9 per month with no volume limit.

SMART monitoring : schedule a quarterly read via CrystalDiskInfo (option "Start with Windows" + "Resident Mode"). The tool sends a Windows notification as soon as a critical attribute degrades. On Mac, DriveDx provides a daemon mode with email alerts. On Linux, smartmontools provides the smartd daemon which can email you the moment an attribute crosses a threshold.

For USB external drives that don't always expose SMART natively, run a manual monthly SMART test: smartctl -t short /dev/sdX then smartctl -l selftest /dev/sdX 5 minutes later.

On SSDs, monitoring logic differs — clicking doesn't exist, but other end-of-life signals appear. See our SSD data recovery and TRIM guide for specifics. And if the problematic drive is an external one that's readable but corrupted (not clicking), our corrupted external drive guide covers chkdsk and recovery software procedures.

10. Quick decision table

This table reflects rates observed in 2023-2024 by public labs (DriveSavers annual stats, Ontrack 2024 report) and remains indicative.

11. After recovery: what to do with the failed drive

Once your data is safely copied to a healthy disk, the failed drive must be handled carefully — not tossed in the household trash. Three options exist depending on the data sensitivity and the level of physical damage:

• Physical destruction at home : for personal drives without ultra-sensitive content, drilling 3 holes through the platters with a 6 mm drill bit renders the data unrecoverable by 99.9 % of methods. Wear safety glasses — the platters are glass on 3.5" Toshiba and HGST models since 2018 and can shatter.
• Professional degaussing : a degaussing service (around $25 per drive) exposes the platters to a magnetic field of 9,000 to 18,000 gauss, erasing all data in under 5 seconds. Mandatory for HIPAA, PCI-DSS or GDPR compliance contexts. Providers: Garner Products, Shred-it, Iron Mountain.
• Certified shredding : for enterprise drives, certified shredders crush the entire drive into 6 mm to 25 mm fragments. NIST SP 800-88 Rev. 1 considers a 6 mm shred to be the gold standard for "destroy" level sanitization.

Never simply reformat a failing drive and resell it on eBay — even a quick format leaves 85 to 95 % of data recoverable by EaseUS in deep scan mode, and the next owner inherits a drive that may fail within 30 days. The 2024 Sims Recycling report estimates that 26 % of second-hand drives sold online still contain recoverable personal data.

Some labs (DriveSavers, Ontrack) offer paid destruction services bundled with successful recoveries: roughly $50 to $150 per drive, with a certificate of destruction valid for compliance audits.

Resources and official sources

• BackBlaze Drive Stats — quarterly reports : failure statistics on 245,000+ drives since 2013.
• Seagate Support — SMART attributes reference : official documentation of SMART attributes on Seagate drives.
• Western Digital Support — SMART troubleshooting : WD guide to interpreting SMART error codes.
• GNU ddrescue manual : official documentation of the cloning tool.
• Our Windows file recovery guide
• Our corrupted external drive guide
• Our EaseUS vs Recuva 2026 comparison