How to compress a PDF without losing quality

Why your PDF is so big

A page of text weighs a few kilobytes. A page that contains a picture can weigh several megabytes, because the picture is stored at its full resolution even when it is printed small on the page. The usual culprits are:

• Scans saved at 300 or 600 DPI, often in color even when the document is black and white.
• Phone photos pasted into a document: a 12-megapixel photo stays 12 megapixels even when it fills a quarter of a page.
• Screenshots and slides exported as large lossless images.
• Hidden extras: page thumbnails, editing data saved by design programs, and metadata.

That is why compressing a PDF is mostly about the pictures. The tool measures how large each image is actually drawn on the page, works out its effective resolution, and scales down only the ones that are sharper than the level you chose.

Which level to choose

DPI (dots per inch) is how many image pixels end up in one inch of the printed page. Fewer pixels mean a smaller file, but also less detail when you zoom in or print. For a picture that fills an A4 page, the levels mean:

Halving the resolution divides the number of pixels by four, which is why going from Print to Ebook makes such a difference. For scanned letters and contracts, Ebook usually looks the same as the original on screen. If the document must be printed by a professional print shop, do not compress it: send the original.

Convert images to grayscale helps less than you might expect: on our color scan of a black-and-white letter it took the Ebook result from 1.3 MB to 1.2 MB, because JPEG already stores color at a lower resolution than brightness. Use it when you want gray pictures, not as a size trick.

What we measured

We compressed three real scans at each level. The letter is the typical case: a document scanned at 300 DPI, where every page is a large photo of paper.

All three files together took 5 to 10 seconds per level. The pamphlet's photos were already below 150 DPI, so Ebook only re-encoded them; Print still saved 30% just by re-encoding at 85% quality. The book scan shows the limit of a careful compressor: its page images are over 60 megapixels and the text layers are JBIG2 masks, so the tool left them alone and returned the file unchanged.

What is changed and what is not

Compression here is careful by design. Photos are re-encoded as JPEG, flat graphics (screenshots, diagrams) are re-compressed losslessly, and the whole file is then rewritten by QPDF, an open-source PDF library compiled to WebAssembly, which packs the internal structure more tightly without changing any content.

Some images are left exactly as they are, because changing them could alter how the page looks:

• images in CMYK or other print color spaces, so colors meant for a press do not shift;
• masks and images with special transparency;
• very small images, where there is nothing to gain;
• images over 60 megapixels, which would use too much memory in a browser;
• images that would get less than 5% smaller.

Text is never turned into pictures, so it stays selectable, searchable and sharp at any zoom. Optional switches also remove page thumbnails and editor data, and the document metadata (title, author, keywords).

Why a PDF does not get smaller

If the result is not smaller than the original, the tool gives you the original back. That happens when:

• the PDF is mostly text, such as a report exported from Word: there are no big pictures to shrink;
• the images are already at or below the chosen resolution: try Screen, or Custom with a lower DPI;
• the images are in CMYK, which is common in files prepared for print;
• the pages are huge single images (over 60 megapixels) or black-and-white JBIG2 masks, as in many book scans from digital libraries;
• the file is encrypted: remove the password first with Unlock PDF (you need to know it), then compress the unlocked copy.

Compressing a PDF for email

Gmail accepts attachments up to 25 MB, and many company mail servers set lower limits. Start with Ebook: in our test it turned a 16.3 MB scanned letter into 1.3 MB. If a document is still too big, try Screen, which is fine for reading on a phone. For very long scans, you can also split the PDF into several attachments.

Other ways to compress a PDF

Windows has no built-in PDF compressor.

How we tested

We ran every test on 24 October 2026 with the tool exactly as published, in Chrome 152 on a desktop PC (AMD Ryzen 7 5700X, NVIDIA GeForce RTX 5060, 32 GB of RAM). Sizes are the ones the tool shows (1 MB = 1,048,576 bytes); times include reading the files.

The letter PDF was built from the five 300 DPI JPEG page scans with each scan placed full-page on A4, which is what a scanner app produces. The page details in the figure were extracted from the compressed PDFs themselves, not simulated.

Test files

• Letter of the Italian Navy, 5 pages scanned at 300 DPI (Letteraspia 1–3, Letterconsole, Lettera ambasc) (public domain · Wikimedia Commons)
• A Letter to Working Mothers, ed. 6 (21-page scanned pamphlet) (public domain · Wikimedia Commons)
• A Letter to A Gentleman in the Country, 1684 (4-page Internet Archive scan) (public domain · Wikimedia Commons)