Source: Backblaze Drive Stats for Q2 2026, published 29 September 2026. Data up to 30 June 2026.
Failure rate of every model
Every model Backblaze runs with more than 100 drives. The lifetime rate covers each model's whole time in service and is the more reliable number. A single quarter of a small model can swing a lot.
| Model | Size | Drives | Avg age | Lifetime failure rate | Apr–Jun 2026 | One of 8 fails in a year | Try it |
|---|---|---|---|---|---|---|---|
| MG11ACA24TE Toshiba | 24 TB | 9,620 | 0.5 yrs | 0.58% | 3.9% | Calculate | |
| WUH721414ALE6L4 WDC | 14 TB | 8,635 | 5.4 yrs | 1.07% | 4.2% | Calculate | |
| WUH721816ALE6L4 WDC | 16 TB | 26,827 | 3.1 yrs | 1.16% | 4.4% | Calculate | |
| WUH722222ALE6L4 WDC | 22 TB | 45,665 | 1.6 yrs | 0.55% | 4.5% | Calculate | |
| HUH721212ALE600 HGST | 12 TB | 2,599 | 6.5 yrs | 1.08% | 5% | Calculate | |
| ST16000NM001G Seagate | 16 TB | 35,100 | 3.6 yrs | 0.65% | 5.3% | Calculate | |
| MG10ACA20TE Toshiba | 20 TB | 25,270 | 1.0 yrs | 1.10% | 6.1% | Calculate | |
| WUH722626ALE6L4 WDC | 26 TB | 7,212 | 0.3 yrs | 0.95% | 6.5% | Calculate | |
| ST12000NM001G Seagate | 12 TB | 13,177 | 5.3 yrs | 1.35% | 7.7% | Calculate | |
| MG07ACA14TA Toshiba | 14 TB | 37,171 | 5.6 yrs | 1.94% | 8.2% | Calculate | |
| MG08ACA16TA Toshiba | 16 TB | 39,932 | 2.8 yrs | 1.01% | 8.2% | Calculate | |
| ST12000NM000J Seagate | 12 TB | 1,079 | 1.8 yrs | 0.00% | 8.5% | Calculate | |
| WUH721816ALE6L0 WDC | 16 TB | 2,970 | 4.4 yrs | 4.07% | 9.2% | Calculate | |
| MG08ACA16TE Toshiba | 16 TB | 6,306 | 4.3 yrs | 1.03% | 9.4% | Calculate | |
| MG09ACA16TE Toshiba | 16 TB | 905 | 0.5 yrs | 1.07% | 9.9% | Calculate | |
| ST14000NM001G Seagate | 14 TB | 10,498 | 5.3 yrs | 1.50% | 11% | Calculate | |
| ST8000DM002 Seagate | 8 TB | 6,903 | 9.6 yrs | 2.46% | 12% | Calculate | |
| MG07ACA14TEY Toshiba | 14 TB | 989 | 3.6 yrs | 2.46% | 13% | Calculate | |
| HUH721212ALE604 HGST | 12 TB | 13,196 | 5.0 yrs | 3.34% | 13% | Calculate | |
| ST8000NM0055 Seagate | 8 TB | 13,112 | 8.7 yrs | 3.44% | 15% | Calculate | |
| ST12000NM0008 Seagate | 12 TB | 18,521 | 6.1 yrs | 3.17% | 16% | Calculate | |
| ST12000NM0007 Seagate | 12 TB | 972 | 6.6 yrs | 4.17% | 17% | Calculate | |
| MG08ACA16TEY Toshiba | 16 TB | 4,857 | 4.5 yrs | 3.66% | 18% | Calculate | |
| HUH721212ALN604 HGST | 12 TB | 9,694 | 7.0 yrs | 7.63% | 18% | Calculate | |
| ST24000NM002H Seagate | 24 TB | 9,631 | 1.0 yrs | 3.30% | 23% | Calculate | |
| ST10000NM0086 Seagate | 10 TB | 965 | 8.5 yrs | 9.33% | 23% | Calculate | |
| ST14000NM0138 Seagate | 14 TB | 1,235 | 5.5 yrs | 8.26% | 38% | Calculate | |
| ST8000NM000A Seagate | 8 TB | 239 | 3.4 yrs | Too few drives | 0.00% | — | Calculate |
| ST14000NM000J Seagate | 14 TB | 497 | 1.4 yrs | Too few drives | 0.00% | — | Calculate |
| ST16000NM000J Seagate | 16 TB | 171 | 1.3 yrs | Too few drives | 2.57% | — | Calculate |
| ST16000NM002J Seagate | 16 TB | 467 | 3.5 yrs | Too few drives | 3.47% | — | Calculate |
Click a column heading to sort. “One of 8 fails in a year” is the chance that at least one drive in an 8-drive array of that model fails within 12 months, using the lifetime rate.
What an annual failure rate means
The annualized failure rate (AFR) is the share of drives you would expect to fail in a year. Backblaze counts every day each drive was running (“drive days”), counts the failures, and scales it to a year:
An AFR of 1.41% means that out of 1,000 drives, about 14 fail in a typical year. For one drive it is a small risk. For an array it adds up: with eight drives at that rate, the chance that at least one fails within a year is 11%. You can try your own numbers in the drive failure probability calculator.
Age matters more than brand or size
Grouping this quarter's data by the average age of each model's drives shows the clearest pattern in the whole set:
| Average drive age | Drives | Failure rate, Apr–Jun 2026 |
|---|---|---|
| Under 2 years | 100,050 | 0.98% |
| 2 to 4 years | 103,554 | 0.95% |
| 4 to 6 years | 98,045 | 2.09% |
| 6 years or more | 52,766 | 4.00% |
Drives under four years old failed at about 1% a year. Models whose drives average six years or more failed at about four times that rate. For a RAID array, this is the number to remember: an array that ran quietly for five years is not proof that it will run quietly for another five.
The same effect hides behind size and brand comparisons. Larger drives look more reliable in this data, but they are also the newest:
| Drive size | Drives | Lifetime failure rate |
|---|---|---|
| 8 and 10 TB | 20,980 | 1.86% |
| 12 TB | 59,238 | 1.92% |
| 14 and 16 TB | 175,425 | 1.04% |
| 20 TB and larger | 97,398 | 0.86% |
The same caution applies to brands: the models in this fleet differ in age, so a brand with older models will show a higher average. Compare models of a similar age, and give little weight to small models with only a few hundred drives.
A second drive failing during a rebuild
A common fear with parity RAID is that a second drive dies while the first one is being rebuilt. Using these failure rates, the chance is small. For an 8-drive array of 16 TB drives, the other seven drives have to survive the rebuild:
| Failure rate of the drives | Rebuild of 30 hours (idle) | Rebuild of 89 hours (busy) |
|---|---|---|
| 0.50% (best models) | 0.01% (1 in 8,300) | 0.04% (1 in 2,800) |
| 1.41% (fleet lifetime) | 0.03% (1 in 2,900) | 0.10% (1 in 990) |
| 3.30% (worst large model) | 0.08% (1 in 1,200) | 0.24% (1 in 420) |
| 7.63% (oldest drives this quarter) | 0.19% (1 in 530) | 0.56% (1 in 180) |
Compare that with the other rebuild risk, an unreadable sector. The same RAID 5 rebuild reads 112 TB from the surviving drives. At the worst-case error rate quoted for NAS drives (1 per 1015 bits), the chance of hitting at least one is 59%. The read error, not a second dead drive, is the main reason to prefer RAID 6 with large drives.
Two caveats make the real risk higher than the table shows. Drives from the same batch tend to fail close together, and a rebuild puts every drive under full load for hours. Both are reasons to keep a spare drive ready and a backup up to date.
About the data
The figures come from Backblaze Drive Stats for Q2 2026. Backblaze, a cloud storage company, has published its drive failure data every quarter for more than ten years. The quarterly table covers models with more than 100 drives; the lifetime table covers models with more than 500 drives and 100,000 drive days, which is why four small models have no lifetime rate here. Boot drives and a few hundred other drives that Backblaze leaves out of its report are not included.
These are data-centre drives running around the clock in a controlled environment. A drive in a home NAS works under different heat, vibration and load, so treat the numbers as a comparison between models, not as a promise for your own drive. Backblaze makes the raw data free to use; this page cites Backblaze as the source and does not sell the data.
Questions
What is a normal hard drive failure rate?
Across Backblaze's fleet, the lifetime rate is 1.41% a year. Good models in their first years run at 0.5–1%; older drives often reach 3% or more.
Which hard drive brand is the most reliable?
The data shows big differences between individual models, not a clear winner by brand. The models also differ in age, which affects failure rates more than the brand does. Compare specific models of similar age.
Do bigger drives fail more often?
Not in this data. The 20 TB and larger drives have the lowest failure rates, but they are also the newest. Their rates will rise as they age.
How often will a drive in my RAID array fail?
Multiply it out: at 1.41% per drive, an 8-drive array has about a 11% chance that at least one drive fails in a year. Use the drive failure probability calculator for your own numbers.