Back up first — before any diagnostic, before any fix attempt. Then match your symptom below for an honest verdict: what is fixable, what needs a recovery service, and what simply needs a replacement drive. Live replacement prices are on this page, refreshed every few hours.
Most of what ranks for "hard drive failing" is written by companies selling data-recovery software, and it shows: the articles end where the software begins, and the one thing recovery software cannot do — fix a physically failing drive — goes unmentioned. This cluster takes the opposite stance, and it is worth stating plainly. Software repairs logical damage: corrupted filesystems, broken partition tables, accidentally deleted files on otherwise healthy hardware. It does not repair heads that click, motors that whine, or platters that have started shedding sectors. When the fault is physical, every minute of additional operation is subtracted from the time you have left to copy data off.
That is why every page in this section opens the same way: back up first. Not after the diagnostic, not after one more reboot to see if the noise stops — first, and in order of importance, so that whatever the drive has left goes to the files that matter. A failing drive that survives long enough to finish a backup has done its job; everything after that is upside.
The second honest stance: we are not a recovery service, and we do not promise recovery outcomes. What we do — the thing the recovery-software sites never do — is serve the step after triage. A large share of people searching these symptoms do not need forensics; they need a replacement drive today, at a fair price, that will not repeat the mistake (wrong recording technology, no warranty, marketplace mystery-meat). The live picks on every page in this cluster are that step: in-stock CMR-family drives, sorted by real $/TB, refreshed on the same cycle as everything on this site.
Live replacement options · prices updated every 4-5 hours · last checked 20 min ago · CMR-family drives only (RAID-safe)
| Drive | Cap | Cond | Warranty | $/TB | Price | |
|---|---|---|---|---|---|---|
| Dell NWCCG 6TB SAS 6G NL Renewed | 6TB | REFURB | see listing | $11.67 | $69.99 | Buy → |
| MDD 6TB SAS 12G 7200RPM | 6TB | NEW | 5 yr | $13.33 | $79.99 | Buy → |
| Dell NWCCG 6TB SAS 6G NL Renewed | 6TB | NEW | see listing | $16.67 | $100.00 | Buy → |
| HGST HUS726060AL5210 6TB SAS 12G | 6TB | REFURB | see listing | $21.50 | $128.99 | Buy → |
Professional recovery exists for one scenario: data that is irreplaceable, unbacked-up, and trapped behind a physical fault. Clean-room labs can swap head assemblies, transplant platters, and image drives that no consumer tool will ever mount again. The honest costs run from hundreds of dollars for straightforward cases into the thousands for complex mechanical work — which is exactly the right price when the alternative is losing years of family photos or a business's only copy of its records, and exactly the wrong price for anything you can re-download, re-rip, or restore from a backup.
Two timing rules make the recovery decision cleanly. First, decide before you experiment: every DIY attempt on a mechanically failing drive — repeated power-ons, long scans, the folklore fixes — reduces what a lab can recover, and labs price accordingly. If the data justifies professional recovery, stop powering the drive and call one. Second, recovery comes before RMA, always: Seagate's own RMA guidance tells customers to recover data before shipping, because warranty replacement drives arrive empty and the original is not returned. A warranty claim on an unrecovered drive is a data-loss event you scheduled yourself.
For everything else — the backed-up array member, the media drive whose contents live on three other disks, the scratch disk — the economics point one way: put the recovery budget toward a better replacement. The picks above show what that money buys today.
Each child page owns one symptom family and ends every section with a verdict, not a maybe. The SMART errors page decodes the attributes that actually predict failure — reallocated sectors, pending sectors, uncorrectable errors, CRC errors, power-on hours — with the defensible thresholds, and calls out the ones that look scary but usually are not. The clicking and not-detected page walks the mechanical and detection symptoms, including the famous 3.3V false alarm that makes perfectly healthy shucked drives play dead. And the failed RAID drive page covers the scenario with a clock on it: a degraded array, where the replacement rules are stricter and the waiting is the risk.
The verdicts use three words consistently. Recoverable-yourself means the fault is logical or trivially external (cable, port, power) and your data is not in danger from the fix. Recovery-service means the fault is physical and the data matters more than the fee. Replace means what it says — and it is the most common verdict, because drives are consumables with a 3-to-5-year service life, and the honest response to most failure symptoms on an aging drive is a calm, well-priced replacement rather than heroics. Our upcoming lifespan guide covers the planning side of that reality.
One more pointer while you are here: if the failing drive is part of a pair or array you are about to refresh entirely, check the upgrade economics tool — in a market where capacities keep getting denser, replacing a failed 8TB with two 8TBs is often worse math than one 16TB or 20TB. And healthy drives you retire still have resale value, which our used-drive tracker prices live.
Every verdict in this cluster gets easier with a current backup, so the discipline deserves its own section. The working rule is 3-2-1: three copies of anything that matters, on two different media, one of them off-site or off-line. It sounds like enterprise ceremony until the day a drive clicks — at which point it converts a crisis into an errand. With a current backup, a failing drive is a shopping decision (the picks above), a warranty claim, or both; without one, the same failure is a recovery-service invoice or a permanent loss, and every triage step happens under pressure.
The economics lean the same way the engineering does. A backup target is one of the cheapest classes of storage we track — refurbished enterprise capacity runs a fraction of new pricing, and a single large drive that holds a second copy of everything costs less than an hour of professional recovery. If this page found you before the failure rather than after it, that is the purchase to make today: pick a capacity above, buy the boring big drive, and schedule the copy. Future-you, reading the clicking-drive section someday, will be a spectator instead of a victim.
No software fixes physical failure. Utilities can repair filesystem-level damage (corrupted tables, bad directory entries) and remap the occasional bad sector, but a drive that clicks, fails SMART thresholds, or accumulates reallocated sectors has a hardware problem that only gets worse. The honest sequence is: back up what you can, then replace.
Back up, immediately, before any diagnostics. Every additional power-on hour on a failing drive is risk, and some diagnostics (long surface scans especially) add hours of sustained load. Copy the irreplaceable data first, in order of importance, then diagnose at leisure.
When the data is worth more than the service costs (typically hundreds to thousands of dollars) and the drive has a physical fault — clicking, dead PCB, no spin-up. Recovery firms work in clean rooms with donor parts and can often recover what software never could. For data that is backed up or replaceable, skip recovery and put the money toward the replacement drive.
If it is inside the warranty period and was not voided (shucked from an enclosure, opened, labels removed), yes — SMART failures and mechanical faults are exactly what warranties cover. Check your serial on the manufacturer's validation page. Note Seagate's guidance: recover your data before starting an RMA, because the replacement arrives empty and your original is not returned.
One historical reallocated sector on an old drive can be watched. A growing count cannot: reallocations that increase across scans mean the surface is deteriorating, and pending sectors (C5) or uncorrectable sectors (C6) at any non-zero value mean plan the replacement now. The distinction is trend versus event.
Unknowable, and planning around an estimate is the classic mistake. Some drives click for weeks; some die the same day. Treat any mechanical symptom as if the drive could fail on the next power cycle: back up now, replace soon, and let the old drive's remaining life be a bonus rather than a plan.
Same capacity or larger, CMR recording (mandatory for RAID and strongly preferred everywhere), and the best live $/TB at your size — which is exactly what the live picks on this page show, refreshed every few hours. If the failed drive lived in a RAID array, see our dedicated RAID replacement guide for the extra rules.