Compress PDF Without Losing Quality: Free 2026 Guide

A PDF that refuses to attach to an email is one of the most mundane and most annoying digital problems there is. The file is not corrupt, nothing is broken. It is just 32 MB when Gmail wants 25, or 6 MB when a government upload form wants 2. The instinct is to reach for an online compressor, hit compress, and hope for the best. But most people never ask the obvious question: what is actually making this file so big in the first place?

Once you understand where the bytes are, compression stops being a gamble. You can tell, before you click anything, whether a PDF will shrink to 10 percent of its size or barely budge, and whether the result will still look right.

What actually makes a PDF big

A PDF is not a single image. It is a container, a small program really, that holds text, fonts, vector art, embedded images and metadata, each stored independently. When you compress a PDF, you are really compressing each of those components separately. Three of them matter, and one is almost always the culprit.

1. Embedded images (usually 80 to 95 percent of the size)

A single scanned page from a home office scanner is stored as an image, often at 600 DPI. That is roughly 3,500 by 4,600 pixels, or about 48 MB of raw pixel data before compression. Drop that into a 10-page PDF and you have a 100 MB file. This is by far the most common reason a PDF is large.

What makes this tricky is that the images are already compressed, usually as JPEG or JBIG2. Recompressing a JPEG is a lossy operation on an already-lossy format. Get it wrong and text edges turn fuzzy, thin lines disappear, and scanned documents become unreadable.

2. Embedded fonts (usually 1 to 5 percent)

A PDF that uses five fonts may embed each one, sometimes the entire font family, sometimes just the glyphs the document actually uses. A full commercial font file is 200 to 500 KB. Subsetting brings that down to 10 to 20 KB, but only if the PDF was exported with subsetting turned on. Word does this by default. LaTeX does not always.

3. Text and vector art (negligible)

This is where the myth dies. Plain text in a PDF is stored as text, actual characters with positioning instructions. A 100-page text-only PDF is typically under 1 MB. If your PDF is 30 MB and it does not contain images, something unusual is going on, usually a multi-megabyte custom font, or an embedded image on the cover page.

The three types of PDF compression

Not all compression is the same. Every compressor you will ever use falls into one of these three buckets, whether it advertises it or not.

Lossless is always safe. Lossy is where things get interesting, and where the reputation of "PDF compressors ruined my document" comes from. The trick is that most scanned documents at 600 DPI carry far more detail than a screen or printer can display. Cutting them to 200 DPI is often visually indistinguishable and cuts file size by 90 percent.

How to compress a PDF (step by step)

Here is a simple workflow that works for any PDF, regardless of what tool you use. It is designed to keep text sharp while cutting the parts that no one will miss.

1. Check the file size. If it is under 5 MB and only text, no compression is needed. Email clients already handle this.
2. Look at what the PDF contains. A 20-page contract with a scanned signature page behaves differently from a 200-page photo book.
3. Choose your target. Most email providers cap at 25 MB. Most government upload forms cap at 2 MB or 5 MB. Aim for one third of the cap.
4. Compress with a browser-based tool that does not upload the file. This keeps confidential documents on your device.
5. Check the result visually. Zoom in on the smallest text you can find. If it is still crisp, you are done.
6. If the result is too big, run it again with a stronger setting. If it looks bad, back off and try a milder preset.

How to compress without losing quality

The phrase "without losing quality" is doing a lot of work. What most people actually mean is: keep the text readable, keep the signatures recognizable, keep the layout intact. Nobody expects a scanned PDF to still show every hair on the paper texture.

Three settings matter most when you are compressing an image-heavy PDF.

Target DPI

• 300 DPI. Print quality. Use this if the PDF will be printed on paper.
• 150 DPI. Screen reading quality. Standard for anything consumed on a laptop or phone.
• 96 DPI. Email preview quality. Acceptable for documents nobody will zoom into.
• 72 DPI. Anything lower starts to lose small text.

For most documents that will be read on a screen, 150 DPI is the sweet spot. You get a 4x reduction in raw pixel count compared to 300 DPI, and the difference is invisible unless you are holding a printout next to the monitor.

JPEG quality (when re-encoding images)

A JPEG at quality 85 is nearly indistinguishable from quality 95 for photographs, and about half the size. Quality 75 starts showing artefacts around sharp edges. Quality 60 is where you begin to see ringing around text on a scanned page.

If you are compressing scans of printed documents, stay at 80 or above. If you are compressing photographs, 85 is fine. Never go below 70 unless the file size is more important than the content.

Convert to grayscale or monochrome for scans

A black-and-white scanned page stored as full-color RGB uses 3x more bytes than it needs to. Converting it to grayscale cuts the size by 66 percent with no visible quality loss on a printed document. Converting to 1-bit monochrome (pure black and white, no gray) cuts it by 90 percent or more, and the result is often sharper than the original.

Why browser-based compression is better than uploading

The traditional way to compress a PDF is to upload it to a server, let the server process it, and download the result. That works, but it means a copy of your document exists on someone else's infrastructure for a period of time. For a restaurant menu, that is fine. For a signed contract, a medical report, or a tax return, it is a real privacy problem.

Modern browser tools, including ToolDeck, do the compression entirely on your device using JavaScript and WebAssembly. The file never leaves your machine. You can verify this by opening your browser's Network tab (Ctrl+Shift+I, then Network) and watching the request log. If you do not see your file being uploaded anywhere, it was not.

This matters even for simple documents because once you upload a PDF to a free service, you have no easy way to know how long it stays on the server, who has access to it, or whether it is being scanned by ad networks or analytics pipelines.

Common mistakes

• Compressing an already-compressed PDF. If someone sent you a 5 MB PDF that was already run through a compressor, running it again will make the images mushy without saving much.
• Using maximum compression on a print document. The result looks fine on screen but reads poorly on paper. If the PDF is going to a printer, use 300 DPI.
• Compressing a text-only PDF. If your PDF is 800 KB and has no images, running it through a compressor will barely change the size. Nothing to compress. Do not waste time on this.
• Splitting and compressing in the wrong order. If you need to send only pages 1 to 5 of a 200-page document, split first, then compress. The compressor will do less work and the result will be smaller.
• Assuming all compressors are equal. Free tools that upload your file often use aggressive downsampling to hit an advertised 90 percent reduction, and the output is unusable. Check the settings.

When not to compress

Some PDFs should not be compressed at all.

• Legal filings where the exact pixel-level content matters (scanned signatures, notarizations, photographs used as evidence).
• PDFs with layered transparency or complex vector art. Compression can flatten layers you wanted to keep.
• PDFs with embedded audio or video, which most compressors ignore but some damage.
• Any PDF you will edit further. Compress the final export, not a working file.

A compressed PDF is not a smaller version of the same file. It is a different file that happens to look similar. Keep the original somewhere safe until you are sure the compressed version is what you want to send.

ToolDeck editorial team