A kilobyte (KB) is a small unit of digital data. Under the international system of units (SI), it is 1,000 bytes. The 1,024-byte unit is properly called a kibibyte (KiB), although some software still labels it KB, a legacy usage that causes confusion. The kilobyte is the first step up from the byte, and the unit directly below the megabyte (MB). Few things are bought by the kilobyte today, but KB figures still turn up in storage pricing rules and technical specifications.
At a glance
- Decimal: 1 KB = 1,000 bytes (10^3). Binary counterpart: 1 kibibyte (KiB) = 1,024 bytes (2^10).
- The gap between the unit and its binary counterpart is about 2.4%, which rarely matters at this scale on its own; some software still labels KiB values as KB.
- As a rough illustration, a short plain-text email or a small text document is typically a few kilobytes.
- For buyers, KB shows up mainly in block sizes, minimum billable object sizes and file size limits.
What problem it solves
Bytes are too small a unit to describe even simple files conveniently, so the kilobyte gives a practical way to state small sizes: configuration files, short documents, log entries, database records and network messages. It also gives storage engineers a unit for describing how storage systems carve up space, since disks, file systems and databases work in fixed-size blocks or pages that are usually a few KB.
The unit’s main source of confusion is historical. Early computers worked in powers of two, and 1,024 was close enough to 1,000 that “kilobyte” was loosely used for both. The International Electrotechnical Commission (IEC) later introduced binary prefixes such as kibi (KiB) so that the 1,024-byte unit has its own name, but the legacy labeling is still common.
How it works
Decimal and binary values. 1 KB = 1,000 bytes in SI usage; 1 KiB = 1,024 bytes. Storage drive makers and most networking standards use the SI units correctly. Some operating systems, notably Windows, calculate sizes in binary units (KiB, MiB, GiB) but label them KB, MB and GB; macOS has used decimal units for many years; Linux tools vary by command and option.
Upper or lower case. The SI symbol for kilo is a lower-case k, so the strictly correct decimal symbol is kB. In practice KB and kB are both used, and neither reliably tells you whether the writer means 1,000 or 1,024 bytes. A lower-case b means bits rather than bytes, so Kb (kilobit) is one-eighth of a KB.
Blocks and pages. File systems allocate space in blocks, and databases store data in pages, commonly a few KB each. A 1-byte file still takes up a whole block on disk, which is why a folder of many tiny files can use noticeably more space than the sum of its file sizes.
Billing minimums. Some cloud object storage tiers, particularly infrequent-access and archive tiers, charge each object as if it were at least a minimum size stated in KB. Metadata overhead per object may also be billed. Workloads that store millions of very small objects can cost more than a per-GB price suggests. See file systems and object storage for how these storage models differ, and our file and object storage page for how buyers compare services.
When it matters for buyers
- Choosing a storage tier for small files. Logs, sensor readings and thumbnails can hit minimum object size rules on cheaper tiers.
- Performance sizing. Block storage performance is often quoted in IOPS at a stated block size (for example 4 KB or 8 KB); the block size behind the figure changes what it means.
- Message and API limits. Email systems, messaging platforms and APIs often cap message or payload sizes in KB.
Questions to ask vendors
- Does this storage tier have a minimum billable object size, and how are smaller objects charged?
- Is per-object metadata or indexing billed separately from capacity?
- At what block size are the quoted IOPS and throughput figures measured?
- When your documentation says KB, do you mean 1,000 or 1,024 bytes?
How it differs from a kibibyte (KiB)
A kibibyte is exactly 1,024 bytes; a kilobyte is 1,000 bytes. Many people and programs still write KB when they mean KiB, which is mislabeling rather than a second definition. The difference is small at this scale, about 2.4%, but it compounds with each step up: by the terabyte the gap is about 10%. The glossary terms Byte and Binary Prefixes cover the base unit and the full set of binary prefixes in more detail.