KB vs KiB: Why Your 1 TB Drive Shows 931 GB
You buy a 1 TB drive, plug it in, and Windows says it's 931 GB. Nothing's broken and nobody stole 69 GB. The drive maker and the operating system are counting with different units that happen to share a name. This guide explains the difference between KB and KiB (and GB and GiB), works out exactly how big the gap is at each size, and covers the other things that genuinely do eat storage space.
Two meanings of "kilo"
In the metric system, kilo means 1,000. A kilometre is 1,000 metres. But computers address memory in powers of two, and 2¹⁰ = 1,024 is close enough to 1,000 that early programmers borrowed the word. For decades, "kilobyte" meant 1,024 bytes in some contexts and 1,000 bytes in others, and the error was small enough that few people cared.
The problem is that the error compounds with each prefix. At kilo it's 2.4%. At giga it's 7.4%. At tera it's almost 10%. In 1998 the International Electrotechnical Commission (IEC) introduced separate binary prefixes to remove the ambiguity:
| Decimal (SI) | Bytes | Binary (IEC) | Bytes | Gap |
|---|---|---|---|---|
| 1 KB (kilobyte) | 1,000 | 1 KiB (kibibyte) | 1,024 | 2.4% |
| 1 MB (megabyte) | 1,000,000 | 1 MiB (mebibyte) | 1,048,576 | 4.9% |
| 1 GB (gigabyte) | 10⁹ | 1 GiB (gibibyte) | 1,073,741,824 | 7.4% |
| 1 TB (terabyte) | 10¹² | 1 TiB (tebibyte) | 1,099,511,627,776 | 10.0% |
Under these definitions, KB, MB, GB and TB are strictly decimal. KiB, MiB, GiB and TiB are binary. The standard is clear. Real-world usage still isn't.
Who uses which
- Drive and USB stick makers use decimal units. A 1 TB drive holds at least 1,000,000,000,000 bytes, and the fine print on the box usually says so.
- Windows File Explorer does binary maths but labels the result with decimal names. It divides by 1,024 and then writes "GB". That's why a 1 TB drive shows as 931 GB.
- macOS (since 10.6 Snow Leopard) and iOS use decimal units. A 1 TB drive shows as about 1 TB, minus what's actually in use.
- RAM is almost always measured in binary units. "16 GB" of RAM is really 16 GiB, because memory chips come in power-of-two sizes.
- Linux tools depend on the flag you pass.
ls -lhanddf -huse powers of 1,024 with short suffixes like G;df -Huses powers of 1,000. - Network speeds are quoted in decimal bits. 100 Mbps means 100,000,000 bits per second.
Where the "missing" space goes
Here's the calculation for a 1 TB drive shown in Windows:
Every byte is still there. It's the same quantity described with a bigger unit. Here's how common sizes come out:
| Advertised | Bytes | Shown in binary units | Apparent shortfall |
|---|---|---|---|
| 128 GB phone | 128 × 10⁹ | 119.2 GiB | 8.8 "GB" |
| 256 GB SSD | 256 × 10⁹ | 238.4 GiB | 17.6 "GB" |
| 512 GB SSD | 512 × 10⁹ | 476.8 GiB | 35.2 "GB" |
| 1 TB drive | 10¹² | 931.3 GiB | 68.7 "GB" |
| 2 TB drive | 2 × 10¹² | 1.82 TiB | 0.18 "TB" |
| 4 TB drive | 4 × 10¹² | 3.64 TiB | 0.36 "TB" |
So on a 4 TB drive, about 360 "GB" seems to vanish, which is more than a whole 256 GB SSD. It's still the same unit confusion, just at a bigger scale.
Space that really is used up
After the unit gap, there's usually some real overhead too. These eat genuine bytes:
- Formatting a drive writes tables that track where files live. NTFS, APFS, ext4 and exFAT all reserve some space for this.
- Windows installs typically create EFI and recovery partitions. Laptops from some manufacturers add their own recovery partition on top.
- A "128 GB" phone ships with the operating system already taking a sizeable chunk, and that's counted as used from day one.
- On Linux, ext4 by default reserves 5% of a partition for the root user. On a 4 TB data drive that's about 200 GB unavailable to regular users. You can lower it with
tune2fs -m. - Files are stored in whole blocks, often 4 KiB. A 1-byte file still uses a 4 KiB block. With hundreds of thousands of tiny files, that adds up.
When the confusion doesn't matter
Here's a reassuring point. If you're working out how many files fit, the answer is the same in either system, as long as you use the same system for both numbers. Say you're filling a 64 GB camera card with RAW photos that are 25 MB each.
Same answer, because both numbers shrank by the same ratio. The trouble starts when you mix systems without noticing. Suppose Windows shows each RAW file as "25 MB", but it's really dividing by 1,024, so each file is 25 MiB, or 26,214,400 bytes. Divide the card's 64 GB by "25 MB" as if both were decimal and you get 2,560. The real figure is 64,000,000,000 ÷ 26,214,400 ≈ 2,441. You'd overestimate by about 119 photos, a 4.9% error that comes straight from the MB versus MiB gap.
In practice, real cards hold a little less than the full count, because the file system takes some space and RAW sizes vary from shot to shot. Treat any such calculation as an upper bound.
Bits, bytes and download times
The same confusion shows up when you estimate downloads, with an extra factor of eight on top. Internet speeds are sold in megabits per second (Mbps). Files are measured in megabytes. One byte is eight bits.
A 4 GB download on a 100 Mbps connection takes at least 320 seconds, about 5 minutes and 20 seconds, assuming the line delivers its full rated speed. Real downloads are slower because of protocol overhead, Wi-Fi conditions and the server's own limits. If your browser shows "11.9 MB/s" on a 100 Mbps line, it's probably using binary MiB and labelling it MB, and you're getting the full speed.
The Data Transfer Rate Converter handles Mbps to MB/s, and the Speed Test shows what your connection actually delivers.
Using the converter
The Digital Storage Converter keeps the two systems separate. KB, MB, GB, TB and PB are decimal (powers of 1,000). KiB, MiB and GiB are binary (powers of 1,024). To check a drive, enter the advertised size in TB and convert to GiB. That's the number Windows will show, labelled "GB".
- Enter the advertised capacity, for example 2 TB.
- Convert to GiB to get the figure Windows will display: 1,862.6.
- If what you see is noticeably lower than that, the difference is real overhead: partitions, file system, or reserved space.
Limitations and edge cases
- Labels lie. Software that writes "MB" might mean either. The converter can only follow the labels you choose. When in doubt, check whether the software's numbers divide neatly by 1,024.
- The converter doesn't know about your file system. It converts units exactly. It can't predict how much space formatting, partitions or reserved blocks will take on a specific drive.
- Some memory cards and older documentation mix units. Floppy disks famously used "1.44 MB" to mean 1,440 × 1,024 bytes, a hybrid of both systems. Expect odd cases in older material.
- The converter covers binary units up to GiB. For TiB, convert to GiB and divide by 1,024.
Once you know the 7.4% rule for GB and the 10% rule for TB, most "missing space" mysteries resolve themselves in a few seconds of mental maths. Anything beyond those percentages is worth investigating.
Check a drive size in seconds
Convert advertised decimal capacity to the binary units your operating system shows, or work out real download times.