How to Check Hard Drive Health: The Complete 2026 Guide From a Data Recovery Lab

To check hard drive health on Windows, open PowerShell as admin and run Get-PhysicalDisk for a quick status, or download CrystalDiskInfo (free) to see all S.M.A.R.T. attributes with color-coded health ratings. On Mac, open Disk Utility and check the S.M.A.R.T. Status field, or install smartmontools via Homebrew and run smartctl -a /dev/disk0. On Linux, run sudo smartctl -a /dev/sda. The five S.M.A.R.T. attributes that actually predict failure are Reallocated Sectors Count, Current Pending Sectors, Uncorrectable Errors, Spin Retry Count, and Command Timeout - if any of these are climbing, back up your data immediately and contact a professional.

But a quick status check only tells part of the story. As data recovery engineers who examine over 1,000 failed drives a year, we see a pattern that frustrates us: people run a quick check, get an "OK" status, and assume their drive is fine. Weeks later, they are sitting in our intake queue with a clicking drive and no backup.

Knowing how to check hard drive health properly means going beyond a single command. It means understanding the physical symptoms a software tool will never detect, reading S.M.A.R.T. attributes the way a recovery technician reads them, and knowing when a drive needs to be powered off immediately rather than scanned one more time.

Most online guides rehash the same two or three methods: run CHKDSK, download a free tool, check the status. That is a start, but it only covers the logical layer. Physical degradation, firmware corruption, and mechanical wear are invisible to those checks until it is too late. At MDrepairs, we open thousands of failed drives every year in our cleanroom using PC-3000 hardware. We have seen thousands of drives that reported "Healthy" right up until total failure.

This guide is different. We wrote it from the bench, not the search results. If you want a fast first look at your drive right now, try our free DIY hard drive diagnostic tool while you read through the rest.

What You Will Learn

• The four layers of hard drive health and why checking only one gives you a false sense of security
• Physical warning signs that no software tool can detect, including the sounds that mean "power off now"
• Step-by-step instructions to check hard drive health on Windows, Mac, and Linux
• How to read S.M.A.R.T. data the way a data recovery engineer reads it - focusing on the five attributes that actually predict failure
• The best hard drive health check tools in 2026, ranked by a lab that uses them daily
• What to do when your drive is failing, including the mistakes that destroy recoverable data

Table of Contents

1. The 10-Minute HDD Health Check (Quick Version)
2. What Is Hard Drive Health (And Why Most Guides Get It Wrong)
3. Warning Signs Your Hard Drive Is Failing
4. How to Check Hard Drive Health on Windows
5. How to Check Hard Drive Health on Mac
6. How to Check Hard Drive Health on Linux
7. How to Read S.M.A.R.T. Data Like a Recovery Engineer
8. Best Hard Drive Health Check Tools (2026)
9. How to Run a Hard Drive Self-Test
10. What to Do When Your Hard Drive Is Failing
11. Inside a Professional Data Recovery Lab
12. How to Prevent Hard Drive Failure
13. Hard Drive Health FAQ

The 10-Minute HDD Health Check (Quick Version)

If you just want an answer right now, this quick HDD health check takes about ten minutes and catches the majority of failing drives. The rest of this guide explains every step in depth.

1. Listen first (1 minute). Power the drive up and listen: clicking, beeping, buzzing, or grinding means a mechanical problem. Stop here, power it down, and skip straight to what to do when your drive is failing. No software test outranks a bad sound.
2. Pull the S.M.A.R.T. overall status (3 minutes). On Windows, run CrystalDiskInfo (the portable version needs no install) and read the Health Status box. On Mac, select the drive in Disk Utility and check the S.M.A.R.T. status line. Anything other than Good or Verified is your answer.
3. Check the five critical attributes (3 minutes). In the attribute table, read the raw values of 05 (Reallocated Sectors Count), 187 (Reported Uncorrectable Errors), 197 (Current Pending Sector Count), 198 (Offline Uncorrectable), and 199 (UDMA CRC Error Count). All zeros is a pass. Any non-zero value on 05, 187, 197, or 198 means the drive is degrading; rising CRC errors on 199 usually mean a bad cable.
4. Sanity-check the temperature (1 minute). Under about 50C is healthy for a hard drive under load. Sustained readings above that shorten drive life and can mimic failure symptoms.
5. Run a short self-test (2 to 5 minutes). Use the drive maker's utility or smartmontools ("smartctl -t short"), then read the result log. A completed test with zero errors is a pass; an aborted or failed test is not.

Scoring your HDD health check: clean on all five steps means the drive is healthy today, so set up monitoring and keep backups. A non-zero critical attribute means back up today and plan the replacement this week. Sounds, a Caution or Bad status, or a failed self-test mean stop using the drive now, and if the data is not backed up, get professional help before it gets worse.

What Is Hard Drive Health (And Why Most Guides Get It Wrong)

Hard drive health is not a single number or a green checkmark. It is the combined condition of four distinct layers inside your drive, and a failure in any one of them can result in total data loss. When you check hard drive health, you need to evaluate all four - not just the one that most tools report on.

The Four Layers of Hard Drive Health

Surface integrity is the condition of the magnetic platters where your data physically lives. Over time, the thin magnetic coating develops weak spots and bad sectors. A few reallocated sectors are normal on an aging drive. Hundreds appearing in a short period are a red flag that the surface is degrading rapidly.

Mechanical health refers to the moving parts: the spindle motor that spins the platters at 5,400 or 7,200 RPM, and the actuator arm that positions the read/write heads nanometers above the surface. Bearing wear, head misalignment, and stiction (when heads stick to the platter surface) are all mechanical failures that no software scan can detect until they cause audible symptoms or total failure.

Firmware stability is the code burned into the drive's controller board and service area on the platters. It manages everything from defect maps to head calibration. Firmware corruption can make a physically healthy drive completely inaccessible. This is one of the most common failure modes we see on Seagate drives from certain model families, and it requires professional tools like the PC-3000 to resolve.

Logical file system health is what most guides focus on exclusively. This is the integrity of the partition table, file system structures (NTFS, APFS, ext4), and directory records. A corrupted file system is often the easiest problem to fix - but only if the underlying hardware is still working. Running repair tools like CHKDSK on a drive with mechanical problems can destroy the remaining data by forcing the failing heads to sweep across damaged platters repeatedly.

Why "OK" Status Lies

Here is a real example from our lab. A client brought in a 2TB Seagate Barracuda that had been reporting "Healthy" in CrystalDiskInfo for over a year. The S.M.A.R.T. status was "Good." Windows showed no errors. The drive passed a quick self-test.

When we opened it in the cleanroom, three of the four read/write heads were scratched and depositing debris across the platters. The drive had been functioning on a single remaining head, reading data from only one platter surface. The client had lost roughly half their files months ago without realizing it - the operating system simply stopped indexing the affected areas silently.

From Our Lab: S.M.A.R.T. status is a threshold-based system. It only flips from "OK" to "Pred Fail" after specific attribute values cross a manufacturer-defined line. We regularly recover drives that reported "OK" the day they arrived at our lab. The status is a lagging indicator, not a real-time assessment. If you are relying on it as your only hard drive health check, you are flying blind.

The takeaway: a proper evaluation of how to check hard drive health must include physical awareness, not just software output. The rest of this guide will show you how to do both.

Warning Signs Your Hard Drive Is Failing

Before you run any diagnostic tool, your drive may already be telling you something is wrong. The earliest signs of hard drive failure are often physical and behavioral, not digital. Learning to recognize them can be the difference between a clean backup and a four-figure recovery bill.

Physical and Audible Symptoms (HDDs)

Mechanical hard drives contain precision components moving at high speeds in a sealed enclosure. When those components begin to fail, they produce sounds. Each sound maps to a specific failure mode, and knowing the difference matters because your response should be different for each one.

Clicking (the "click of death"). A repetitive click-click-click pattern, sometimes fast and sometimes slow, almost always means the read/write heads are failing. The actuator arm swings the heads into position, they fail to read the servo tracks on the platter, and the drive resets them to the parking ramp and tries again. Every retry cycle risks physical contact between the heads and the platter surface. A hard drive clicking like this needs to be powered off immediately. Each additional power cycle increases the chance of platter scoring, which can make data permanently unrecoverable.

Grinding or scraping. This is worse than clicking. A grinding sound typically means the heads have already made contact with the spinning platter and are scoring the magnetic surface. This is an active, ongoing destruction of your data. We have seen clients lose entire drives because they "wanted to copy just a few more files" while the drive was grinding. Do not do this. Power off. Unplug. Do not plug it back in.

Beeping on power-up. A beep (not from the computer's speaker, but from the drive itself) usually indicates motor seizure or stiction. The platters are not spinning. This can happen after a drop, after a long period of inactivity, or due to bearing failure. The drive may have recoverable data on the platters, but it needs a motor swap or head-platter separation in a cleanroom to access it.

Buzzing or whining. A continuous buzz often means the spindle motor is struggling but still partially functional. Bearing wear is the most common cause. The drive may still be accessible intermittently, but it is on borrowed time. Back up everything you can reach without delay.

From Our Lab: Out of the last 200 clicking drives we received at MDRepairs, 73% had platter damage visible under our inspection microscope. The clients who powered the drive off after the first click had significantly better recovery outcomes than those who kept trying to copy files. If you hear any of these sounds, the single best decision you can make is to stop using the drive. See our full breakdown in our hard drive failure guide.

Silent Performance Symptoms

Not every failing drive announces itself with noise. Some drives degrade quietly, and the symptoms mimic software problems so closely that many users blame Windows, malware, or a bad update instead of the hardware underneath.

Extreme slowness during file operations. If copying a folder that used to take seconds now takes minutes, or if your computer freezes for 10-30 seconds when opening a file, the drive may be encountering bad sectors and retrying reads. The operating system waits patiently while the drive's firmware attempts to read the same sector dozens of times before giving up. You experience this as a freeze.

Files disappearing or becoming corrupted. You saved a document yesterday and today it will not open, or it opens with garbled content. Folders you know existed are suddenly empty. This can be a logical file system issue (fixable) or a surface degradation issue (much more serious). The key distinction: if the corruption is spreading - more files affected over days and weeks - it is almost certainly physical.

Blue screens (BSOD) referencing disk errors. Stop codes like KERNEL_DATA_INPAGE_ERROR, CRITICAL_PROCESS_DIED, or NTFS_FILE_SYSTEM can all point to a hard drive failing beneath the operating system. If these blue screens happen repeatedly and especially during file access rather than random moments, suspect the drive.

Unusually long boot times. A computer that used to reach the desktop in 30 seconds now takes 3-5 minutes. The drive is struggling to read the OS files. You might see the manufacturer logo screen for an abnormally long time, or Windows may appear to hang on the loading animation.

The challenge with these silent symptoms is distinguishing logical problems from physical ones. A rough test: if the issue gets progressively worse over time rather than appearing suddenly after a software change or update, lean toward hardware. If the drive also feels unusually warm to the touch or makes any subtle vibration changes, hardware becomes even more likely.

SSD-Specific Warning Signs

Solid-state drives fail differently from mechanical drives. There are no moving parts, so there are no sounds. But SSDs have their own distinctive failure patterns that are worth knowing, especially since SSDs are now the majority of drives in laptops and desktops.

Read-only mode. Many SSDs are designed to enter a read-only state when the NAND flash cells have exhausted their write endurance. You can still access your files, but you cannot save anything new. If you suddenly cannot save files to your SSD and get write errors, the drive may have entered this protective mode. This is actually one of the better failure outcomes - your data is still accessible, so back it up immediately before anything else goes wrong.

Drive disappears after boot. The computer starts up, the SSD works for a few minutes, then it vanishes from the system. You open File Explorer or Finder and the drive is gone. A reboot brings it back temporarily. This pattern often indicates controller firmware instability or failing NAND blocks in the firmware area. It tends to get worse until the drive stops appearing entirely.

Wrong capacity or "SATAFIRM S11" in device name. If your 1TB SSD suddenly reports as 8MB, or if the drive name changes to something like "SATAFIRM S11" (common on certain SanDisk and WD drives), the firmware has crashed. The data is usually still on the NAND chips, but accessing it requires professional tools that can read the raw flash memory and reconstruct the file system. This is not a DIY fix.

For a deeper look at SSD-specific problems and recovery options, see our SSD data recovery service page or our detailed guide on how to recover data from an SSD.

From Our Lab: SSD failures are increasing in our intake as flash storage ages. In 2025, SSDs accounted for roughly 35% of the drives we received - up from about 20% two years earlier. The good news is that SSD data is often fully recoverable when the controller fails, because the NAND chips themselves are usually intact. The bad news is that recovery requires chip-off reading or specialized hardware like the PC-3000 SSD module, which means it is not something you can do at home.

How to Check Hard Drive Health on Windows

Windows gives you several built-in ways to check hard drive health, plus excellent free tools that go much deeper. We will start with the fastest method and work up to the most thorough. If you are here because your drive is already making noises or acting erratically, skip the software checks entirely and go straight to our warning signs section - running diagnostics on a physically failing drive can cause additional damage.

Method 1 - WMIC / PowerShell Command (30-Second Quick Check)

This is the fastest way to check hard drive health Windows users have available without installing anything. It queries the drive's S.M.A.R.T. system for a simple pass/fail status. It takes 30 seconds and requires no downloads.

Step 1: Right-click the Start button and select Terminal (Admin) or Windows PowerShell (Admin). You need administrator privileges to query hardware directly.

Step 2: Type the following command and press Enter:

wmic diskdrive get status

You will see either OK or Pred Fail for each connected drive. "OK" means the S.M.A.R.T. thresholds have not been crossed. "Pred Fail" means at least one critical attribute has exceeded its threshold and the drive's own firmware is predicting imminent failure.

Important note on WMIC deprecation: Microsoft has been deprecating WMIC since Windows 11. It still works on most systems in 2026, but the modern replacement is PowerShell's built-in cmdlet. Run this instead for more detailed output:

Get-PhysicalDisk | Select FriendlyName, MediaType, OperationalStatus, HealthStatus

This command returns the drive name, whether it is an HDD or SSD, its operational status, and its health status. You will see "Healthy" or "Warning" or "Unhealthy" in the HealthStatus column. The MediaType column is useful if you have multiple drives and are not sure which is your SSD and which is your HDD.

From Our Lab: An "OK" or "Healthy" result from these commands is not a clean bill of health. It only means the drive has not crossed its own failure thresholds - thresholds that manufacturers set conservatively to reduce warranty claims, not to protect your data. We have recovered data from drives that reported "OK" literally the same day the client brought them to us. Use this as a quick sanity check, not a diagnosis. If the result is "Pred Fail" or "Unhealthy," that is definitive - the drive is telling you it is dying, and you should stop using it for anything except making a backup immediately.

Method 2 - CrystalDiskInfo (Recommended Free Tool)

CrystalDiskInfo is the single best free tool for checking hard drive health on Windows. It reads the full S.M.A.R.T. attribute table and presents it in a clean, color-coded interface. This is the tool we recommend to every client who wants to monitor their drives at home.

Step 1: Download CrystalDiskInfo from crystalmark.info. Use the "Standard Edition" installer or the portable ZIP if you prefer not to install anything.

Step 2: Run the program. It will immediately scan all connected drives and display a dashboard. Each drive gets a colored health indicator at the top.

Step 3: Interpret the color codes:

Step 4: Look beyond the color. The real value of CrystalDiskInfo is in the attribute table below the health indicator. Scroll down and focus on the Raw Values column (the rightmost one). The "Current" and "Worst" columns are normalized values that manufacturers calculate differently - the raw value tells the actual count.

The attributes that matter most are covered in detail in our S.M.A.R.T. data section below, but here is a quick reference for what to watch:

• Reallocated Sectors Count (05): Any raw value above 0 means the drive has already found bad sectors and moved data elsewhere. A handful is normal on older drives. A number that keeps increasing is a serious warning.
• Current Pending Sector Count (C5): Sectors the drive suspects are bad but has not yet confirmed. These are sectors waiting to be retested on the next write operation.
• Uncorrectable Sector Count (C6): Sectors the drive tried to read, failed, and could not recover. Data in these sectors is lost at the drive level.
• Power-On Hours (09): Total hours the drive has been running. Not a health metric per se, but context for the other values. A drive with 50,000 hours and rising reallocated sectors is on a very different trajectory than one with 2,000 hours and the same count.

Set CrystalDiskInfo to run at startup (Function menu > Resident) and it will sit in your system tray, monitoring your drives continuously. It can alert you when a status changes from Good to Caution - giving you advance warning before a full failure.

Method 3 - Windows Event Viewer (Hidden Disk Errors Most People Miss)

This is the method that most hard drive health guides completely skip, and it is one of the most valuable. Windows logs every disk-related error in the Event Viewer, and these logs can reveal problems that S.M.A.R.T. has not flagged yet. If you want to check hard drive health on Windows more thoroughly than 99% of users, this is how.

Step 1: Press Win + R, type eventvwr.msc, and press Enter to open Event Viewer.

Step 2: In the left panel, expand Windows Logs and click on System.

Step 3: Click Filter Current Log in the right panel. In the Event Sources dropdown, check the following sources:

• disk - low-level disk driver errors
• ntfs - file system structure errors
• volsnap - volume shadow copy errors (often triggered by bad sectors)

Click OK to apply the filter.

Step 4: Review the filtered results. Here are the event IDs that should concern you:

What to look for: The pattern matters more than individual events. Open your filtered log and look at the timestamps. If you see disk errors clustering in the past few days or weeks after a long period with none, the drive is actively deteriorating. A handful of errors spread across years of operation is normal wear. Twenty errors in the past week is a drive calling for help.

From Our Lab: Event Viewer is one of the first things we check when a client brings in a drive that "just stopped working." In many cases, Windows had been logging disk errors for weeks or months before the failure. The client never noticed because they never looked. We recommend checking Event Viewer once a month as part of a basic hard drive health check routine. It takes two minutes and can give you weeks of advance warning that S.M.A.R.T. alone would miss.

Between these three methods - the quick PowerShell check, CrystalDiskInfo for ongoing monitoring, and Event Viewer for historical error patterns - you have a thorough approach to checking hard drive health on Windows that covers far more ground than any single tool. If all three methods look clean and you are not experiencing any of the warning signs listed above, your drive is in genuinely good shape.

How to Check Hard Drive Health on Mac

Mac users have fewer built-in diagnostic options than Windows users, and the one they do have - Disk Utility - is widely misunderstood. Here is how to check hard drive health properly on macOS, starting with the built-in tool and then moving to the professional-grade method that actual technicians use.

Disk Utility First Aid (What It Actually Does - and Doesn't)

Step 1: Open Finder > Applications > Utilities > Disk Utility.

Step 2: Select your drive in the left sidebar. Make sure you select the physical drive (the top-level entry), not a volume or partition underneath it.

Step 3: Click First Aid, then click Run.

First Aid will scan the file system structure and repair minor inconsistencies - things like incorrect file counts, orphaned directory entries, or catalog file errors. On APFS volumes, it also checks the container structure and snapshot integrity.

Here is the critical point that Apple's documentation buries: First Aid checks the file system, not the hardware. It does not read S.M.A.R.T. data. It does not scan the platter surface for bad sectors. It does not evaluate mechanical health. A drive can pass First Aid with flying colors and have heads that are about to fail.

Worse, running First Aid on a drive with physical problems can accelerate the damage. First Aid reads every directory structure on the disk. If the heads are weak or the surface is degrading, you are forcing the drive to perform thousands of read operations across the entire platter surface. We have received drives at MDRepairs where the client ran First Aid "to check if the drive was okay" and the process pushed a marginal drive into total failure.

Rule of thumb: If the drive is making any unusual sounds, if it is extremely slow, or if files have been disappearing, do not run First Aid. It is a file system tool, not a diagnostic tool. For more on what Disk Utility can and cannot do, see Apple's Disk Utility support page.

Using smartmontools on macOS (The Pro Method)

If you want to read actual SMART data on a Mac - the same data that data recovery engineers use to evaluate drive health - you need smartmontools. This is an open-source package that queries the drive's S.M.A.R.T. system directly and returns the full attribute table, error log, and self-test history.

Step 1: Install Homebrew if you have not already. Open Terminal (Applications > Utilities > Terminal) and paste:

/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

Step 2: Install smartmontools:

brew install smartmontools

Step 3: Identify your drive's device path. Run:

diskutil list

Your main drive is usually /dev/disk0. External drives will be /dev/disk1, /dev/disk2, etc. Note the device path for the drive you want to check.

Step 4: Run the full S.M.A.R.T. report:

smartctl -a /dev/disk0

This produces a detailed output with several sections. Here is what to focus on:

• SMART overall-health self-assessment test result: This is the pass/fail at the top. "PASSED" means no attributes have crossed failure thresholds. "FAILED" means at least one has - treat this as an emergency.
• SMART Attributes table: The same attribute data you would see in CrystalDiskInfo, but in text form. Focus on the RAW_VALUE column for attributes 05 (Reallocated Sector Count), C5 (Current Pending Sector), and C6 (Uncorrectable Sector Count).
• SMART Error Log: A record of recent read/write errors. Any entries here indicate the drive has encountered problems it could not silently correct.
• SMART Self-test log: Results of any self-tests the drive has run. Look for entries showing "Completed without error" vs. those showing failures at specific LBA addresses.

For external drives connected via USB, you may need to add the -d sat flag to pass S.M.A.R.T. commands through the USB bridge:

smartctl -a -d sat /dev/disk2

Not all USB enclosures support S.M.A.R.T. passthrough. If you get errors, try connecting the drive directly via SATA or Thunderbolt if possible. For more details on smartmontools, visit smartmontools.org.

From Our Lab: We use smartmontools on every Mac and Linux system in our lab. It gives us the same raw data as our PC-3000 hardware for the S.M.A.R.T. layer - free and in seconds. If you own a Mac and you care about your data, installing smartmontools is a five-minute investment that lets you check hard drive health at the same depth as a professional. Pair it with a monthly check and you will catch degradation long before it becomes a crisis. For the drives beyond software diagnostics, that is where our hard drive data recovery service picks up.

How to Check Hard Drive Health on Linux

Linux gives you the most direct, unfiltered access to your hard drive's health data. No wrappers, no dumbed-down interfaces - just raw diagnostic power through the terminal. If you are comfortable with the command line, Linux is arguably the best platform to check hard drive health because smartctl was originally built for Unix-based systems.

Even if Linux is not your daily driver, booting from a live USB (like Ubuntu) is one of the best ways to diagnose a drive that Windows or macOS refuses to recognize. We use this approach in our lab regularly when a client's OS drive will not boot.

smartctl Commands for Linux

The smartmontools package includes smartctl, the same underlying tool that powers most GUI disk health apps on every platform. Install it with your package manager:

sudo apt install smartmontools    # Debian, Ubuntu, Mint
sudo yum install smartmontools    # CentOS, RHEL, Fedora (older)
sudo dnf install smartmontools    # Fedora (newer)

Once installed, identify your drives. The lsblk command gives you a clean tree view of all block devices, while sudo fdisk -l provides more detail including partition tables and drive sizes:

lsblk
sudo fdisk -l

With your drive identifier in hand (typically /dev/sda for the first SATA drive), pull the full SMART report:

sudo smartctl -a /dev/sda

This single command outputs everything: drive model, serial number, firmware version, power-on hours, temperature, and every SMART attribute with raw and normalized values. If SMART monitoring is disabled on the drive (rare, but it happens), enable it first:

sudo smartctl -s on /dev/sda

For NVMe drives, the device path changes slightly. Use /dev/nvme0 or /dev/nvme0n1 instead. The smartctl tool handles NVMe natively in recent versions, so the same -a flag works. For older installations, you may need to update smartmontools from smartmontools.org to get NVMe support.

GNOME Disks (GUI Method)

Not everyone wants to live in the terminal, and that is perfectly fine. If you are running Ubuntu, Fedora, or any GNOME-based desktop, the built-in Disks application provides a clean graphical interface to SMART data.

Open the Disks app from your application menu. Select your drive from the left sidebar. Click the three-dot menu in the top-right corner and choose "SMART Data & Self-Tests." You will see a health indicator at the top - a green "Disk is OK" or a red warning - along with a scrollable table of every SMART attribute.

The visual health indicator is helpful for a quick glance, but do not rely on it exclusively. GNOME Disks uses a simple threshold check, which means a drive can show "OK" while harboring hundreds of reallocated sectors that have not yet crossed the manufacturer's failure threshold. Always scroll down and review the raw values, particularly for the five critical attributes we cover in the next section.

From Our Lab: About 30% of the drives we receive for hard drive recovery were still reporting "Healthy" in their SMART status when the owner last checked. The overall status is a lagging indicator. Individual attribute trends tell the real story.

How to Read S.M.A.R.T. Data Like a Recovery Engineer

Knowing how to check hard drive health is only half the battle. The other half is understanding what the numbers actually mean. SMART (Self-Monitoring, Analysis, and Reporting Technology) tracks dozens of attributes, but most of them are informational - power-on hours, start/stop counts, temperature readings. They tell you about the drive's life, not its remaining lifespan.

Recovery engineers focus on a handful of attributes that directly correlate with imminent failure. After examining thousands of failing drives in our lab, these are the five SMART attributes that matter most.

The 5 Critical SMART Attributes That Predict Failure

The most important thing to understand about these SMART attributes is that a rising trend matters far more than any single reading. A drive with 3 reallocated sectors that has stayed at 3 for two years is in a very different situation than a drive that went from 0 to 3 in the past week. The second drive is actively failing.

Check your SMART data monthly, and compare it to previous readings. Tools like CrystalDiskInfo and Hard Disk Sentinel can track these values over time and alert you when any attribute starts climbing.

How to Read Raw Values vs Normalized Values

Every SMART attribute has two values that confuse most people: the normalized value and the raw value. Understanding the difference is critical if you want to accurately assess your drive's health.

Normalized values are scaled from 0 to 253 (sometimes 0 to 100, depending on the manufacturer). Each attribute also has a threshold value set by the manufacturer. When the normalized value drops below the threshold, the drive is officially "failing" according to the manufacturer's criteria. Think of it like a fuel gauge - it tells you the general zone, but not the exact number of gallons.

Raw values are the actual event counts. This is the number that matters. Raw value 5 for Reallocated Sectors Count literally means five sectors have been remapped. Raw value 47,832 for Power-On Hours means the drive has been running for about 5.5 years.

The problem with relying on normalized values is that manufacturers set their thresholds very conservatively - often designed to avoid warranty claims rather than protect your data. A drive might have a threshold of 36 for reallocated sectors, which means it will not report "failing" until the normalized value drops below 36, potentially allowing hundreds or thousands of bad sectors before the status changes.

From Our Lab: We regularly see drives arrive at our facility with a normalized SMART status of "Good" but 2,000 or more raw reallocated sectors. The drive firmware considers itself healthy. Your data would disagree. Always look at raw values, not normalized scores, when you check hard drive health.

In smartctl output, raw values appear in the rightmost column labeled "RAW_VALUE." In CrystalDiskInfo, hover over any attribute to see both normalized and raw values. In DriveDx on macOS, raw values are displayed alongside a visual health bar for each attribute.

SSD-Specific Health Metrics

Solid-state drives do not have spinning platters or read/write heads, so the SMART attributes that predict HDD failure (reallocated sectors, uncorrectable errors) are less relevant. SSDs have their own set of health indicators tied to the nature of NAND flash memory.

Percentage Used is the single most important SSD health metric. It represents how much of the drive's rated write endurance has been consumed. At 100%, the drive has reached its rated lifespan - it may continue working, but the manufacturer no longer guarantees reliability. Most SSDs display this as attribute 173, 177, or a vendor-specific ID.

Total Host Writes (sometimes reported as Total LBAs Written) tells you exactly how much data has been written to the drive over its lifetime. Compare this to the manufacturer's rated TBW (Total Bytes Written) to calculate remaining endurance. A Samsung 870 EVO 1TB, for example, is rated for 600 TBW.

Media and Data Integrity Errors are especially important for NVMe drives. Any non-zero value here means the drive has detected data corruption that its internal error correction could not fix. This is the NVMe equivalent of uncorrectable errors on an HDD.

Wear Leveling Count indicates how evenly the drive has distributed writes across all NAND cells. A low wear leveling count means the drive is doing its job well. An abnormally high count can indicate the controller is struggling to find healthy cells to write to.

If your SSD is showing concerning metrics, our guides on how to recover data from an SSD and NVMe SSD data recovery explain your options in detail. SSD failures are fundamentally different from HDD failures and often require different recovery approaches.

Use our free interactive diagnostic tool to check these SMART values on your drive and get a plain-English assessment of what they mean for your data.

Best Hard Drive Health Check Tools (2026)

You do not need to memorize SMART attribute IDs to keep tabs on your drive's health. The right tool translates raw data into actionable information. Here are the best options available in 2026, from quick visual checks to deep diagnostic suites.

CrystalDiskInfo is the go-to for most Windows users. It launches instantly, color-codes your drive's health status (blue for good, yellow for caution, red for bad), and displays every SMART attribute in a clean table. Set it to run at startup and it will sit in your system tray, alerting you when any attribute crosses a threshold. It is the fastest way to check hard drive health on a Windows machine.

smartmontools is the Swiss Army knife of drive diagnostics. The smartctl command-line tool runs on every major operating system and provides the most complete SMART data output of any tool. It also lets you run short and long self-tests directly from the terminal. If you only learn one diagnostic tool, make it this one. Download it from smartmontools.org.

DriveDx fills a gap on macOS where built-in tools fall short. It provides real-time drive monitoring, historical trend tracking, and plain-English health assessments for both HDDs and SSDs. The $24.99 price is worth it if you rely on a Mac for critical work.

Hard Disk Sentinel excels at continuous monitoring. It calculates an overall health percentage and estimated remaining lifespan based on the rate of SMART attribute degradation. Its alert system can notify you by email, sound, or even trigger a backup script when a drive starts showing signs of failure.

SeaTools, Samsung Magician, and WD Dashboard are manufacturer-specific utilities. They are free, well-designed, and optimized for their respective brands. If you have a Seagate, Samsung, or Western Digital drive, install the matching tool for firmware updates and brand-specific diagnostic features alongside a general-purpose tool like CrystalDiskInfo.

MDRepairs DIY Diagnostic Tool takes a different approach. Instead of dumping raw SMART data on your screen, it walks you through a guided diagnostic process - asking about symptoms, noises, and behavior - and gives you a clear assessment of whether your data is at risk and what to do next. It works in any browser on any device, requires no installation, and was built by our recovery engineers based on the same diagnostic logic we use in the lab. If you are not sure how to interpret SMART numbers, start here.

How to Run a Hard Drive Self-Test

SMART monitoring is passive - it records events as they happen. A hard drive self-test is active. It forces the drive to scan its own platters, test its electronics, and report back on what it finds. Running periodic self-tests is one of the most reliable ways to catch problems before they cause data loss.

Short Test vs Long Test

There are two types of self-tests, and both are built into the drive's firmware. They are read-only and completely non-destructive - no data is modified or erased during testing. You can safely run them on any healthy drive.

Short test: Takes 2 to 5 minutes. It checks the drive's electrical and mechanical components, reads a small sample of sectors, and verifies the servo positioning system. Think of it as a quick checkup.

sudo smartctl -t short /dev/sda

Long test: Takes 1 to 4 hours depending on drive capacity. It performs a complete sequential read of every sector on every platter. This is the definitive surface scan - if there is a bad sector anywhere on the drive, the long test will find it.

sudo smartctl -t long /dev/sda

Both tests run in the background. The drive continues to function normally during testing, although performance may be slightly reduced. Check the results after the estimated completion time:

sudo smartctl -l selftest /dev/sda

The output shows a history of all self-tests with their results. You want to see "Completed without error" for every entry. Any other result - "Read failure," "Aborted by host," or a specific LBA address where the test failed - indicates a problem that needs attention.

On Windows, you can run the same tests through CrystalDiskInfo or the command-line version of smartctl included with smartmontools. On macOS, DriveDx provides a GUI interface for launching and monitoring self-tests.

When Self-Tests Are Dangerous

Self-tests are safe for healthy drives. They are not safe for drives that are already physically failing. This distinction is critical and can mean the difference between a successful recovery and permanent data loss.

If your drive is making clicking, grinding, or beeping sounds, do not run a self-test. These sounds indicate mechanical failure - damaged read/write heads, a seized spindle motor, or head-platter contact. A self-test forces the drive to move its heads across every sector, which can cause a failing head assembly to scratch the platter surfaces and destroy data that was previously recoverable.

If your drive is intermittently detected - appearing and disappearing from your system, or taking minutes to show up after power-on - do not run tests. This behavior usually indicates failing electronics or a head that is on the verge of complete failure. Any additional stress can push it over the edge.

If your drive shows any of these symptoms, skip the self-test and go straight to a professional data recovery evaluation.

Warning: Never confuse self-tests with destructive testing tools. Commands like badblocks -w perform a read-write test that overwrites every sector on the drive, permanently destroying all data. The -w flag means "write mode." We have received drives in our lab where the owner ran badblocks -w thinking it was a diagnostic - it erased everything. If you are running any command on a drive with important data, verify that it is read-only before executing.

What to Do When Your Hard Drive Is Failing

You have learned how to check hard drive health. Now the results are bad. Maybe SMART is throwing warnings, maybe the drive is making noises it never made before, or maybe files are disappearing. What you do in the next few minutes matters more than anything else in determining whether your data survives.

Most data loss is not caused by the initial failure. It is caused by what the user does after the failure. Every action you take on a failing drive - every program you open, every scan you run - risks making the situation worse. Here is exactly what to do, in order.

The First 5 Minutes (Emergency Action Plan)

1. Stop using the drive immediately. Close all programs. Do not save files, do not open new ones, do not try "one more thing." If the failing drive is your boot drive, shut down the computer. A clean shutdown is better than a hard power-off, but do not wait - shut down now.
2. Do NOT run chkdsk, Disk Utility First Aid, or any repair utility. These tools attempt to fix filesystem corruption by writing to the drive. On a physically failing drive, writes can accelerate damage. Chkdsk in particular can delete file entries it considers damaged, making recovery harder even after the physical issue is resolved.
3. Do NOT open the hard drive. Hard drives operate in a sealed environment with filtered air. Opening the enclosure outside of a certified cleanroom exposes the platters to dust particles that are larger than the head-platter gap. A single fingerprint on a platter can destroy data across thousands of sectors. No matter what you saw on YouTube, do not open it.
4. Document the symptoms. Write down exactly what happened: what sounds the drive makes, what error messages appeared, whether the drive is detected in BIOS, and when the problem started. This information helps recovery engineers diagnose the issue faster, which can save you both time and money.
5. Disconnect the drive and contact a professional. Remove the drive from your computer (or simply leave the computer powered off). Reach out to a professional data recovery lab for an evaluation. Most reputable labs, including ours, provide diagnostics.

If the Drive Still Works - Safe Data Backup

Not every failing drive needs emergency treatment. If your SMART data shows rising reallocated sectors but the drive is still functioning - no strange noises, no freezing, files still accessible - you have a window of opportunity to back up your data before the situation gets worse.

The safest approach is a sector-by-sector clone using ddrescue, a Linux tool designed specifically for copying data from failing drives. Unlike a normal file copy, ddrescue skips bad sectors on the first pass, recovers the easy data first, and then goes back to retry the problem areas. This minimizes stress on the failing regions of the drive.

sudo ddrescue /dev/sda /dev/sdb rescue.log

If you are not comfortable with command-line cloning tools, our DIY diagnostic tool can help you assess the risk level and determine whether a self-backup is safe or whether you should go straight to professional recovery.

When is cloning safe? When SMART values are degrading but the drive is responsive, quiet, and accessible. When is it too risky? When the drive makes any abnormal noise, freezes during access, takes a long time to be recognized, or has already lost sectors containing critical files. In those cases, every power-on cycle risks further damage. Send it to a lab.

Wondering about costs? Our guide on how much data recovery costs breaks down pricing by failure type so you know what to expect before reaching out.

Inside a Professional Data Recovery Lab

There comes a point where no software tool, no terminal command, and no amount of Googling can get your data back. When the drive clicks, when firmware is corrupted, when heads have crashed into the platter surface - consumer tools hit a wall. This is where professional data recovery begins.

What Happens When Software Cannot Help

The centerpiece of any serious data recovery lab is the PC-3000, a hardware-software platform manufactured by ACE Laboratory. This system communicates directly with the hard drive's firmware in what is called "technological mode" - bypassing the drive's normal operating logic entirely.

Why does this matter? Because many drive failures are not mechanical at all. They are firmware failures - corrupted module tables, damaged translator maps, or corrupted SMART data that causes the drive to refuse to spin up. The PC-3000 can read, edit, and rewrite firmware modules to bring a non-functional drive back to a state where data can be imaged.

For drives with partial head failure, the PC-3000 can disable individual failing heads and image data only from the platters served by the remaining good heads. If a drive has four heads and one has crashed, we can often recover 75% or more of the data by reading from the three surviving heads. This is impossible with consumer software because the drive's own controller does not expose head-level control.

The imaging process itself is a sector-by-sector clone that bypasses the operating system entirely. The PC-3000 reads raw sectors from the drive and writes them to a healthy destination drive, building a map of good and bad sectors as it goes. Recovery engineers can then work on the cloned image without touching the original, preserving the source drive for additional passes if needed.

From Our Lab: About 40% of the drives we recover have firmware-level corruption that makes them invisible to any computer, even though the mechanical components are perfectly fine. These drives are often misdiagnosed as "dead" by IT shops that lack firmware-level tools. If your drive is not detected but is not making unusual noises, firmware corruption is a strong possibility.

The MDRepairs Recovery Process

We built our process to be straightforward. No confusing tiers, no hidden fees, no pressure tactics. Here is how it works from start to finish:

1. Mail in or drop off your device. We accept drives from anywhere in the United States. Our mail-in process includes a prepaid shipping label so you can send your drive to our lab at no cost.
2. Diagnostic evaluation. Our engineers examine your drive to identify the exact failure - mechanical, firmware, logical, or electrical. We determine what is recoverable and what is not. The phone consult is always free, and the diagnostic fee is credited toward your recovery if you proceed.
3. Get a firm quote. Based on the diagnostic findings, we provide a fixed price quote. No surprises, no escalating costs mid-recovery. And our guarantee is simple: if we cannot recover your data, you do not pay. No Data, No Charge.
4. Lab recovery in our certified cleanroom. All physical work - head swaps, platter transplants, component-level repair - happens inside our ISO-5 certified cleanroom under laminar flow hoods that filter particles down to 0.3 microns.
5. Data returned on a new healthy drive. Once recovery is complete, we transfer your files to a new external drive and ship it back to you with a full file listing so you can verify everything before we securely erase the original.

Get a free quote to find out what recovery would cost for your specific situation.

Take a look inside our lab and see how we handle drive recoveries from start to finish:

How to Prevent Hard Drive Failure

The best data recovery is the one you never need. Hard drives are mechanical devices with a finite hard drive lifespan, but proper care and monitoring can extend that lifespan significantly and give you enough warning to back up before a failure becomes catastrophic.

The 3-2-1 Backup Rule

This rule has been the gold standard for data protection for decades, and it remains the simplest framework that actually works:

• 3 copies of your data (the original plus two backups)
• 2 different media types (for example, an internal drive and an external drive, or a local drive and cloud storage)
• 1 offsite copy (cloud storage, a drive stored at another location, or a safe deposit box)

Why three copies? Because any single storage device can fail at any time. Two copies protect you against a single failure. Three copies, on two different media types with one offsite, protect you against a single failure, a local disaster (fire, flood, theft), and the unlikely but possible scenario of two simultaneous failures.

Cloud backup services like Backblaze, iDrive, or even Google Drive make the offsite copy effortless. For critical business data, test your backups quarterly by actually restoring files - a backup you have never tested is a backup you cannot trust.

Schedule Regular SMART Monitoring

Checking your drive's health once is good. Checking it automatically on a schedule is better. Knowing how to check hard drive health once is useful, but setting up continuous monitoring means you get warned at the first sign of trouble without lifting a finger.

Windows: Install CrystalDiskInfo and enable its "Resident" (startup) mode. It will run quietly in your system tray and display a warning balloon notification the moment any SMART attribute crosses a threshold. This is the easiest set-and-forget option.

Mac and Linux: Configure the smartd daemon that comes with smartmontools. Edit /etc/smartd.conf to monitor all drives and send email alerts when attributes change. A basic configuration looks like this:

/dev/sda -a -o on -S on -s (S/../.././02|L/../../6/03) -m your@email.com

This monitors all attributes, runs a short self-test daily at 2 AM, runs a long self-test every Saturday at 3 AM, and emails you if anything looks wrong. At minimum, you should be reviewing SMART data monthly - but automated monitoring means you will never miss a warning.

Environmental Factors That Kill Drives

Temperature is the most common environmental killer. Hard drives should operate below 45 degrees Celsius (113 degrees Fahrenheit). Sustained temperatures above 50 degrees Celsius accelerate lubricant breakdown in the spindle bearing and can cause thermal expansion of the platters, leading to read errors. Ensure adequate airflow in your case and clean dust filters regularly.

Vibration is especially dangerous for HDDs. The read/write heads float nanometers above the platter surface - any vibration can cause momentary contact. Keep HDDs away from speakers, subwoofers, and other sources of low-frequency vibration. In multi-drive systems, use rubber mounting grommets to isolate drives from each other's vibrations.

Power instability causes more drive failures than most people realize. A sudden power loss during a write operation can corrupt filesystem metadata, the drive's internal translator table, or even damage the motor controller electronics. Use a UPS (uninterruptible power supply) for any system with important data. Even a basic $50 UPS provides enough battery time for a clean shutdown during an outage.

For a deeper look at what causes drives to fail, read our detailed guide on hard drive failure causes and warning signs.

Hard Drive Health FAQ

Take Action Before It Is Too Late

Now you know how to check hard drive health on every major operating system, how to read the SMART attributes that actually matter, and what to do when the numbers start going in the wrong direction. The most important takeaway is this: learning how to check hard drive health is the easy part - the hard part is doing it consistently. Check regularly, learn the warning signs, and act fast when something looks wrong. Every hour a failing drive runs is an hour of additional risk to your data.

If your drive is healthy, set up automated monitoring and follow the 3-2-1 backup rule. If your SMART data is showing concerning trends, back up now - not tomorrow, not this weekend, now. And if your drive is already clicking, grinding, or undetected, stop powering it on and let a professional handle it.

Here is what to do next:

• Check your drive right now with our free DIY diagnostic tool - no downloads, works in your browser, and gives you a clear health assessment in minutes.
• Already seeing warning signs? Get a free quote from our recovery engineers. We will tell you exactly what is going on and what it will take to recover your data.
• Need to send in a drive? Mail in your device with our prepaid shipping label. Diagnostics, fast turnaround, and our no data, no charge guarantee means you have nothing to lose.

At MDRepairs, we have recovered data from thousands of drives that owners thought were beyond saving. Whether you are dealing with a drive that is just starting to show bad sectors or one that will not power on at all, we have the tools and the expertise to get your data back. No Data, No Charge - that is our guarantee.

When to Stop Checking and Protect the Data

A health check is useful only while the drive can still respond normally. If SMART values are worsening, the drive clicks, files copy slowly, or the system freezes during access, do not keep testing just to get one more confirmation. At that point, every read attempt can add stress and reduce what can be recovered.

Stop especially if CHKDSK, Disk Utility, or recovery software hangs, reports I/O errors, or makes the drive disconnect. Power the computer down, remove the drive if you can do so safely, and keep it off. MDRepairs accepts drives from anywhere in the US by insured mail-in service. You can send the drive to our lab for a $100 diagnostic, applied toward recovery. We will explain what failed before paid recovery work starts, and if there is no data, there is no charge.

Every Way a Failing Drive Gets Recovered

Clicking, not spinning, not detected, or dropped after a fall, each failure needs a different bench process, from cleanroom head swaps to firmware repair on PC-3000 hardware. If the steps in this guide did not bring the drive back, the fix is physical, not software. See the full range of data recovery services we handle in the lab, or start an insured mail-in case. The diagnostic is $100, applied toward recovery. No data, no charge.