Digital & media

MB, GB, MiB and GiB Explained

Why does my 1 TB drive only show 931 GB?

You buy a 1 TB drive. Windows says 931 GB. Nothing is broken, nobody is cheating you, and the missing 69 GB was never there — because the two numbers are measuring the same bytes with two different definitions of the prefix "giga".

Two meanings of one prefix

In the SI system, prefixes are decimal. Kilo means 1,000. Mega means a million. Giga means a billion. This is true for grams, metres, watts and everything else, and storage manufacturers use it for bytes as well.

Computers, however, address memory in powers of two. The nearest power of two to 1,000 is 1,024 (2¹⁰), and early computing simply borrowed the prefix names for those values. So "kilobyte" came to mean 1,024 bytes in some contexts and 1,000 in others — the same word, two values, 2.4% apart.

The gap compounds with each step:

Decimal (SI) Bytes Binary (IEC) Bytes Gap
1 kB1,0001 KiB1,0242.4%
1 MB1,000,0001 MiB1,048,5764.9%
1 GB1,000,000,0001 GiB1,073,741,8247.4%
1 TB10¹²1 TiB2⁴⁰10.0%
1 PB10¹⁵1 PiB2⁵⁰12.6%

The IEC fixed this in 1998 — sort of

In December 1998 the IEC approved a set of binary prefixes as an international standard: kibi (Ki), mebi (Mi), gibi (Gi), tebi (Ti) and so on, each a contraction of "kilo binary", "mega binary" and the rest. NIST documents them alongside the SI prefixes. Under this scheme the ambiguity disappears entirely: GB always means 10⁹ bytes and GiB always means 2³⁰.

The standard is unambiguous. Adoption is not. Roughly where things stand:

  • Storage manufacturers — decimal. A 1 TB drive is 10¹² bytes.
  • Networking — decimal throughout, for both speeds and volumes.
  • macOS — decimal since OS X 10.6 (2009). Reports a 1 TB drive as 1 TB.
  • Linux — mixed; most tools offer both, and many default to binary with IEC labels.
  • Windows — binary values with decimal labels. Reports 2³⁰ bytes as "1 GB".
  • RAM — genuinely binary everywhere, and labelled GB by universal convention.

Windows is the source of nearly every "where did my storage go" question, because it is the one place that displays a binary number under a decimal name. It is not wrong about the byte count — it is only wrong about the label.

Working through the 1 TB drive

  1. The drive contains 1,000,000,000,000 bytes. That is what the box means by 1 TB.
  2. Windows divides by 1,024 three times: 1,000,000,000,000 ÷ 1,073,741,824 = 931.32.
  3. It displays "931 GB". The correct label would be 931 GiB.
  4. Formatting then takes a little more for the file system's own structures — typically well under 1% for NTFS on a drive this size.

So of the apparent shortfall, almost all of it is labelling and only a sliver is real overhead.

Why RAM is different

Memory is addressed by binary address lines: n address bits reach exactly 2ⁿ locations. There is no way to build a memory module with 1,000,000,000 addressable bytes that is not wasteful, so RAM capacities are genuinely powers of two. A "16 GB" stick holds 17,179,869,184 bytes — it is 16 GiB, and everyone calls it GB.

Storage has no such constraint. A drive is a sequence of sectors and can be any size the manufacturer likes, so there was never a reason for it to be binary. That is the underlying reason the two conventions diverged in the first place — it is not arbitrary.

Which should you use?

  • Storage capacity, downloads, data caps, network transfers — decimal. That is what manufacturers, carriers and standards all mean.
  • RAM, memory allocation, page and block sizes — binary, because the hardware is.
  • When it matters — write GiB explicitly, or give the byte count. Ambiguity that costs 10% is worth four extra characters.

Don't forget bits

One more distinction sits on top of all this: bytes versus bits. Storage is in bytes (capital B); network speed is in bits (lower-case b). A 100 Mbps connection moves 12.5 MB/s, an eightfold difference that has nothing to do with decimal versus binary. That one has its own guide, because it causes just as much confusion on its own.

Stack the two together and you can be out by a factor of eight and another 7.4% simultaneously — which is roughly how a "gigabit" connection ends up feeling nothing like a gigabyte per second.

Try it in the converter

Sources

Every reference below was opened and checked against the live page on the date shown. Where this guide states a definition, it comes from one of these documents rather than from a secondary summary.

  1. Definitions of the SI units: The binary prefixes NIST · link checked 12 August 2026
  2. SI Brochure — The International System of Units (SI), 9th edition BIPM (Bureau International des Poids et Mesures) · link checked 12 August 2026

More guides

Browse all guides →

Found an error in this guide, or a definition that has moved on? Tell us — corrections are made against the published standard and the update date above is changed when they are. See also how conversions are calculated.