AZW3 - Amazon Kindle Format 8

AZW3 is the file extension Amazon uses for Kindle Format 8, usually shortened to KF8. It arrived with the Kindle Fire as the replacement for MOBI, the ageing format the first Kindles had inherited from Mobipocket. The change was mostly about typography: publishers wanted real fonts, styled headings and page layouts that survive the trip to a device, and MOBI could not carry them. In practice an AZW3 is close to an EPUB 3 book repacked in a Kindle wrapper, which is roughly what Amazon's conversion tools do when a publisher uploads one.

How an AZW3 file is built

The outer shell is the same Palm Database container MOBI used, so the file is binary and the same tools recognise the family. What changed is the content inside: instead of nineties-era HTML, KF8 holds a compiled subset of HTML5 and CSS3. That is what lets an AZW3 carry embedded fonts, drop caps, nested and numbered lists, background images, tables, fixed-layout pages and styling rules that respond to screen size.

Text is still cut into records and compressed with the PalmDOC or HuffCDIC schemes, both of them lossless. The markup is stored as a skeleton plus fragments, so a reader can render part of a chapter without parsing all of it. The practical consequence is that you cannot pull tidy HTML files out of an AZW3 the way you can out of an EPUB. Books bought from Amazon usually arrive with DRM, which encrypts the content and ties it to the buying account, and many titles are now delivered in Amazon's newer KFX format instead.

What metadata an AZW3 carries

AZW3 inherits MOBI's EXTH header, short for extended header: a list of entries each tagged with a number instead of a name. The fields you meet are the updated title, author, publisher, description, ISBN, subject tags, publication date, rights, language, contributor and the ASIN, the identifier Amazon gives a title in its catalogue. A separate entry flags which record holds the cover image. The Palm header underneath contributes the internal database name, normally a squashed form of the title, plus creation and modification timestamps.

Two entries are worth knowing about. The contributor field is almost never a human being: conversion tools write themselves into it, so it typically reads as Calibre or kindlegen with a version number, revealing what produced the file. And the content type code, cdeType, separates a shop purchase, marked EBOK, from a personal document you emailed or sideloaded yourself, marked PDOC. That code, together with an ASIN, says where the file came from. So the personal traces in an AZW3 are the author and contributor names, the producing software and its version, the timestamps and the store identifier. A cover image inside the file keeps its own EXIF data, the camera information block stored in photographs, which can include a device model and even GPS coordinates. What is not in the file is your reading history: highlights, bookmarks and last page read are kept by the device and the account, not written back into the book.

When to use AZW3

AZW3 is the format for reading on Kindle hardware and Kindle apps with the layout the publisher intended, and it sideloads over USB without a shop involved. Against MOBI the case is easy: the same container, far better typography, and the only thing you give up is the very oldest Kindles, which cannot read it. Against EPUB it is a question of reach. EPUB is opened by a wide range of devices, apps and library systems, while AZW3 is mostly Kindle territory. If a book has to work everywhere, keep the EPUB; if it lives on a Kindle, AZW3 is the native fit.