How to read CBR and CBZ comics on a Kindle

A CBZ is a zip file full of page images, and a CBR is the same idea inside a RAR archive. Neither is a book format: there is no page order beyond the file names, no cover flag, no reading direction, and no page size. Kindle firmware reads Amazon's own formats, PDF, and a handful of text formats, and has no code for looking inside an archive. For a longer look at what these archives are and which apps open them on a computer or phone, see the companion guide at /guides/cbz-cbr-files-explained.

Copying a CBZ over USB is a dead end; the file sits in storage and never appears on the home screen. Renaming it to .zip or .epub changes nothing, because the reader checks the contents, not the name, and Send to Kindle refuses archives for the same reason. The pages have to be repackaged as a document type the Kindle understands, and in 2026 the sensible target is EPUB.

Every current Kindle has a 300 pixels-per-inch panel, but 300 ppi is a density, not a size. What matters for a comic page is the pixel count of the screen: 1072 by 1448 on the 6-inch basic Kindle, 1236 by 1648 on the 6.8-inch Paperwhite, 1264 by 1680 on the 7-inch Oasis, Colorsoft, and 2024 Paperwhite, and 1860 by 2480 on the 10.2-inch Scribe. A page with at least that many pixels fills the screen sharply; a smaller one is shown smaller, or enlarged and softened.

The Device menu sets the box every page is scaled to fit inside, and pages are only ever scaled down, so a low-resolution scan stays that way whichever profile you choose. The 6-inch model from 2022 onward matches the Kindle (11th gen) profile, and the 2024 7-inch Paperwhite has its own profile, Kindle Paperwhite (12th gen, 2024), at 1264x1680. Colorsoft is the one Kindle profile that keeps color; its color layer resolves 150 ppi, so colored areas look softer than line art.

E-ink reflects room light, and its brightest white is closer to newsprint than to a backlit screen. Anything not pure white in the source shows as gray, and scans are rarely pure white: aged paper photographs as warm gray and faded ink as dark gray. Midtones fare worst, because the panel compresses them and the page reads flat.

For every grayscale Kindle profile the converter reduces each page to luminance, then stretches the brightness histogram so the 1st percentile becomes black and the 99th becomes white, which turns yellowed paper back into paper. Pages that already span the full range are left alone, as are nearly uniform ones such as a black endpaper. A gamma of 1.8 then darkens midtones to offset the panel's washout, and the page is dithered down to the 16 gray levels e-ink can physically show, so halftones and gradients do not band. On the Colorsoft profile none of this runs.

The EPUB holds one JPEG per page at the profile's resolution. A dithered grayscale page at Paperwhite size typically weighs 200 to 500 KB, so a 200-page volume lands between 40 and 100 MB depending on the art. Scribe pages carry roughly twice the pixels and produce files about twice as large; Colorsoft color pages are larger again.

Measure that against the two limits: 200 MB per file through the apps and web uploader, 50 MB per email. Email suits a single issue, not a manga volume, so default to the app or web page. A book over 200 MB is almost always a Scribe conversion of a long volume; use a smaller profile or split the archive before converting.

CBR means the pages are inside a RAR archive rather than a zip. RAR is proprietary, so it never became a standard library everywhere the way zip did. Many converters skip it, and those that support it often rely on extraction code written for the older RAR4 layout. RAR5, introduced in 2013, changed the header format, so a RAR4-only tool reports a RAR5 file as empty or corrupt even though the archive is fine.

The converters linked below bundle a WebAssembly build of unrar that reads both RAR4 and RAR5 inside the browser, so a CBR needs no extra software. The container is detected from the first bytes rather than the extension, since plenty of files named .cbr are really zips. Not handled: password-protected archives, and multi-volume sets split into part files, which need combining into one archive first.