File signature index
111 distinct magic numbers, one page each, grouped by what they identify.
A file signature is the fixed run of bytes a format writes at a known offset, usually right at the start. It is what a program checks when it wants to know what a file really is, because a filename can say anything. This index has one page per distinct signature rather than one per format, so the bytes that DOCX, XLSX, EPUB, APK and JAR all share are explained once, in the place where that fact belongs.
Which page do you want?
If you have a hex string and want to know what it is, use the wide magic-numbers table: it is sortable, filterable and downloadable as CSV and JSON. If you already know the format and want the bytes explained, pick it from the list below. If you have the actual file, the in-browser identifier reads its first bytes locally and tells you, without uploading anything.
What each page tells you
Every signature page gives the hex, the offset the bytes sit at, what they read as in ASCII where that is meaningful, and a byte-by-byte breakdown. It names the formats that share the signature, including ones outside this database, so you know when the bytes narrow a file down rather than prove it. Of the 111 signatures here, 11 sit at an offset other than zero and 3 have a documented trailer at the end of the file as well, which is how a truncated download gives itself away.
Image
Video
Audio
Document
Archive
System
Data
Disk
Font
3D
Design
eBook
Frequently asked questions
What is a file signature?
A file signature, also called a magic number, is a fixed run of bytes that a format writes at a known offset, usually the very first bytes of the file. Programs read it to decide what a file actually is, because the extension in the filename can say anything.
Why is there one page per signature instead of one per format?
Because many formats share the same bytes. Everything built on ZIP, including DOCX, XLSX, PPTX, EPUB, APK and JAR, starts with 50 4B 03 04. Explaining that once, on the ZIP signature page, is accurate; repeating it per format would claim those bytes prove something they do not.
Do all file formats have a signature?
No. Plain-text formats such as CSV, JSON, YAML and most source code have no fixed header, and neither do some headerless binary formats. Those are identified by their structure and their extension instead, and they are deliberately left out of this index rather than guessed at.
Can I use this signature data in my own project?
Yes. The whole dataset is available as JSON from the free API with no key and no rate limit, licensed CC BY 4.0. Attribution to WhatFileType is the only condition.