Like many things, every hard drive stores a "diary" of its own deterioration. Hard drives keep track of their performance and problems through SMART (Self Monitoring Analysis Reporting Technology). The tools to read it are free and take two minutes. However, most user guides fail to provide any information about the actual parameters that indicate a potential problem. They typically list dozens of SMART attributes but never explain which ones are significant enough to indicate an impending failure. Moreover, few guides go into detail about how to respond when any of these attributes show movement.
Here are some common SMART attributes that can indicate potential problems with your drive. There are others, but these are the most commonly referenced: Reallocated Sectors, Pending Sectors, Uncorrectable Sectors, Spin Retries and the trend over time (did any of them show any movement since last month?).
While there are several ways to view your SMART data on a PC running Windows, we highly recommend downloading CrystalDiskInfo. It’s free, small, and widely regarded as the de facto tool to use for viewing SMART data. Once installed, simply click on any local disk (not networked disks) and you’ll see a simple table showing each attribute along with a coloured verdict across the top: Blue = Good, Yellow = Caution, Red = Bad. On a Mac, SMART data can be accessed via Apple’s Disk Utility app. Open Disk Utility and select any internal disk drive. Under the drive’s info is a field labelled ‘SMART Status’, which shows either a word that indicates overall health (‘Verified’), or a warning (‘Failing’, etc.) indicating potential issues with the drive. For those who feel comfortable working in Terminal (and want to get detailed SMART data), you can run the smartctl command via Homebrew’s smartmontools package. On any platform, accessing SMART data via your drive manufacturer’s utility (Seagate’s SeaTools or WD’s Dashboard) yields similar results.
One honest caveat before you trust any of them on an external drive: some USB enclosures do not pass SMART data through to your computer so if there is no value or “unsupported” on that drive, it does not mean healthy. It means that the SMART data is unreadable via that cable.
Find these rows in the attribute table. Their raw values should be zero and remain zero.
Reallocated Sectors (05): patches of the disk that failed and were replaced with spares. The single most-watched number by the trade: a handful are survivable, but a climbing count of this number indicates that the drive is announcing its retirement. Current Pending Sectors (C5): sectors that the drive can no longer read reliably and is waiting for final judgement on; they typically contain your files that are now unreadable. Any non-zero value here will require you to back up immediately. Uncorrectable Sectors (C6): sectors that, even after all retries, failed to successfully read: confirmed damage. Spin Retry Count (0A): the motor required more than one attempt to spin up; while far less common, this mechanical red flag is far more serious than it looks. Fifth, the trend: take a screenshot today and compare next month. A drive that has three reallocated sectors forever is tired; a drive that goes from zero to three to eleven is failing on a pretty specific schedule.
Power-on hours and temperature give context (a 5-year old drive with 40,000 power-on hours deserves way more scepticism) but neither predicts anything on its own.
When a drive indicates it’s being cautious, don’t immediately think to do a series of long-running scans (a complete surface test, CHKDSK, etc.). It takes hours to perform each deep scan on a piece of hardware that has already indicated that it is working hard to provide access to its contents. Pending sectors tend to grow under exactly that type of sustained reading. Therefore, copy your irreplaceable files off first (prioritising the most critical folders), then determine what you are going to do with the drive. If the file copying process begins to hang your computer, if the drive makes new noises (clicking, buzzing, spinning down repeatedly), stop immediately and power off the drive. At this point, every additional minute spent attempting to continue running the drive will cost you data. In addition, clicking-drive recovery on a bench, imaging data directly from the drive independently of the host computer, is safer than continuing to use the failing drive. A drive that has already failed (not detected by software, incorrect size, or hanging indefinitely) should skip the checklist and proceed directly into a professional hard drive recovery service.
With regard to external drives, there is one key thing that complicates the “check” function. Many enclosures contain a USB-to-SATA bridge that does not allow SMART command output. As such, your check-function shows no health data at all. Do not assume this means the drive has passed inspection. This simply represents a communication barrier between the enclosure/bridge combination and your diagnostic software. There are three ways to get around this limitation in increasing levels of effort: Try a diagnostic tool that supports the USB bridge in your enclosure (CrystalDiskInfo can read more types of enclosures than most other tools). Try placing the drive in a different docking station or caddy. For a drive containing valuable information that is behaving poorly, send the drive out for native diagnosis on a diagnostic bench. There is also a consideration unique to external drives. Due to their smaller form factor, external enclosures often produce more audible output than larger internal drives. Although a well-behaving external drive may be perceived as louder than an internally mounted drive, it is the rhythm and pattern of sounds made by the drive that indicate potential problems, rather than pure volume.
No, and most people do not know this. The CHKDSK utility works through checking the file system’s logical structure, not the drive’s physical health. It writes corrections where needed. If your disk is physically deteriorating, then running CHKDSK repeatedly is equivalent to applying a stress test to your drive. Reading your SMART data provides you with the disk’s own record of how well it has been performing and how much life it has left. SMART data is collected automatically from the disk itself.
Treat percentages as marketing and look beyond them. While a drive may appear to be “95 per cent healthy”, the number of reallocated sectors within the drive could still be increasing week after week. The rate of increase in the number of reallocated sectors is often used as a strong indicator of impending failure. Beyond the headline percentage value, there are several other values associated with disk performance which engineers consider when assessing whether a disk is fit for use: the five attributes above. Seeing those at zero matters far more than any headline score.
Clicking sounds coming from your disk whenever it tries to access data. This is the best evidence of impending failure. SMART reporting gives information about what electronic components have reported back to the operating system; however, mechanical issues inside a hard drive can cause failures which are invisible to SMART reporting, i.e., damage to the magnetic head(s) of the disk. Thus, if you hear clicking sounds (or buzzing, or continuous spin up/spin down cycles) and your SMART scores suggest otherwise, power the drive down and assume that it is failing.
Do not judge a drive based on a single metric. Disk temperatures can vary greatly depending on conditions, and drives are comfortable well into the 40s Celsius; e.g., during a hot summer day, even a normally reliable drive may operate slightly warmer than normal. What matters is whether your drive operates within its expected range, especially whether it increases in temperature significantly compared to past behaviour. Some common examples include a drive suddenly running 15C hotter than it did previously, or a drive running consistently above 55C. Both of these conditions would require investigation as potentially symptomatic of either excessive loading causing excessive wear/heat, or potentially due to overheating caused by faulty cooling mechanisms such as failed fans. Log the temperature readings along with each run of your SMART testing tool so that ‘running hotter’ becomes a measurable quantity and not simply subjective experience.
Free diagnostic service at our lab: if you'd like to bring your drive in person to our lab in Belfast for review, you are welcome to do so without charge or obligation, and you get a written quote before any recovery work.